{ "data": { "a": { "slug": "deepseek-harness", "name": "DeepSeek Harness (dsh)", "vendor": "DeepSeek", "vendorUrl": "https://deepseek-harness.github.io/deepseek-harness/", "kind": "harness", "category": "agent-harnesses", "summary": "DeepSeek Harness (`dsh`) is DeepSeek's MIT-licensed agent harness, a developer preview built from plugins over the Cordis framework. It runs as a local Web UI, a one-shot headless command, an ACP server or a JSON-RPC server for its Python SDK.", "url": "https://www.anchorterminal.com/tools/deepseek-harness", "markdownUrl": "https://www.anchorterminal.com/tools/deepseek-harness.md", "slimMarkdownUrl": "https://www.anchorterminal.com/tools/deepseek-harness.min.md", "jsonUrl": "https://www.anchorterminal.com/api/v1/tools/deepseek-harness.json", "dns": "deepseek-harness.grade.anchorterminal.com", "repo": "https://github.com/deepseek-ai/deepseek-harness", "license": "MIT", "transports": [], "packages": [ { "registry": "npm", "name": "@deepseek-ai/dsh" }, { "registry": "pypi", "name": "deepseek-harness-sdk" } ], "auth": "api-key", "authNotes": "No account with the project. A model key goes in the environment, `$DSH_HOME/.credentials.yaml`, a `.env` file or the Web UI settings, for DeepSeek (`DEEPSEEK_API_KEY`), a catalogue provider or a custom OpenAI or Anthropic-compatible endpoint, including a self-hosted one. The Web UI listens on loopback and authenticates each request.", "pricing": "free", "pricingNotes": "Free and MIT, with nothing to buy for the harness. The owner pays the model provider, usually DeepSeek's API at its published per-token rates, or nothing with a self-hosted model.", "priceSummary": "Free · OSS", "where": "local", "x402": { "level": "no", "evidence": "No x402, MPP or L402 in the README, the CLI reference or the docs index (checked 2026-10-18).", "endpoints": [] }, "toolCount": null, "popularity": { "githubStars": null, "npmWeekly": 244843, "pypiWeekly": null, "asOf": "2026-10-18" }, "docsUrl": "https://deepseek-harness.github.io/deepseek-harness/", "llmsTxt": "https://deepseek-harness.github.io/deepseek-harness/llms.txt", "capabilities": [ "agent.harness", "agent.mcp-client", "agent.multi-agent" ], "tags": [ "official", "harness", "coding-agent", "cli", "open-source", "typescript", "python", "mcp", "acp", "llms-txt", "telemetry-default-on", "pre-1.0", "preview", "free", "no-card", "local" ], "lastRelease": "2026-10-17", "graded": true, "anchor": { "graded": true, "score": 48.5, "grade": "D", "agentReady": false, "rank": 1123, "ranked": true, "rankOf": 1510, "categoryRank": 45, "methodology": "0.7", "run": "2026-10-01", "scores": { "ergonomics": 68, "maintenance": 71, "payments": 60, "reliability": 38, "schema": 77, "security": 40, "transparency": 53 }, "pending": [ "performance", "tasks" ], "assessment": { "confidence": "medium", "date": "2026-11-04" }, "negative": -8, "negativeNotes": [ "2026-09-08. CVE-2026-82533 (CVSS 4.0 9.4 Critical, published by VulnCheck), also GitHub-reviewed as GHSA-8m2g-8cgm-3vcp for `@deepseek-ai/dsh-client-connection`. The local HTTP agent-control API trusted a client-supplied loopback Host header, so a sandboxed tool process could reach it, escape the sandbox and turn off approvals, and a tunnelled port exposed it remotely. Fixed in 0.1.2-alpha.1 with commit 3e24087 (25 August 2026), which every later version carries. A third-party Critical, fixed, under two months old, -2. https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/82xxx/CVE-2026-82533.json", "2026-11-04. CVE-2026-82033 (CVSS 4.0 8.6 High, published by VulnCheck). The web profile's HTTP and WebSocket RPC control plane had no authentication before 0.1.2-alpha.2, so any local process, including one run by another OS user, could call session methods, run the agent's bash and file tools as the victim and approve tool calls. Fixed in 0.1.2-alpha.2 (30 August 2026). The record lists commit 3e24087, the fix for CVE-2026-82533, among the affected code, so this is a second flaw and not a second record of the first. A third-party High, fixed, -2. https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/82xxx/CVE-2026-82033.json", "2026-10-16. CVE-2026-95182 (untrusted search path leading to code execution, CISA-ADP CVSS 3.1 10.0) and CVE-2026-95183 (prompt injection through context compaction, CISA-ADP 8.8), both assigned by MITRE against commit c291e796 (10 September 2026, in 0.1.6-alpha.1 and later) and resting on third-party write-ups we didn't read. No fix is known. Counted one by one as third-party Critical and High records, -2 each. https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/95xxx/CVE-2026-95182.json", "2026-09-28 to 2026-10-26. Eight VulDB records, each saying the vendor was contacted early and did not respond, are counted at nothing under this run's reading of third-party records (Mediums uncounted, Lows free). CVE-2026-101078, CVE-2026-101102, CVE-2026-108788 and CVE-2026-108789 (CVSS 4.0 Medium), CVE-2026-101131 and CVE-2026-101132 (28 September and 12 October), and CVE-2026-116367 (sandbox createContext, to 0.1.0-rc.5) and CVE-2026-116368 (browser auth improper authorization, 0.2.0-rc.0 to rc.2, CVSS 4.0 6.9 Medium, 3.1 7.3) of 26 October. None of them lists a fixed version. https://github.com/CVEProject/cvelistV5/blob/main/cves/2026/116xxx/CVE-2026-116368.json" ], "verdict": "Headless runs, ACP and a Python SDK make it easy to drive from a script, and approvals fail closed when nobody can answer. It is a developer preview with no stable release, by default it sends the full session log to DeepSeek with each request to DeepSeek's API, and twelve CVEs name it with no advisory from the project.", "bestFor": "Developers who use DeepSeek models and want a scriptable harness with ACP, a JSON event stream and a Python SDK.", "strengths": [ "MIT licence, with headless `--json`, an ACP stdio server and a JSON-RPC server for the Python SDK from one `dsh` command", "In headless and ACP runs an approval nobody answers is denied, so the default `workspace-write` plus `ask` preset fails closed", "File writes are confined to the workspace by bwrap or Landlock on Linux, Seatbelt on macOS and restricted tokens on Windows", "`--dump-config-schema` prints a JSON Schema 2020-12 document for the composed configuration, and the docs serve llms.txt and Markdown", "38 `dsh-v` tags between 17 August and 4 November 2026, with dated upgrade guides naming breaking changes for each pre-release since 29 September" ], "weaknesses": [ "By default each request to DeepSeek's official API carries the session log, including message text, tool arguments and results, up to 8 MiB a request", "Developer preview with no stable release. npm `latest` was 0.2.0-rc.2 on 18 October, and there is no changelog, only per-version upgrade guides", "The sandbox confines writes only. Reads and network access are not confined, and `web_fetch` runs without approval", "No SECURITY.md or security.txt. Twelve CVEs from September to November 2026 name it, eight saying the vendor did not respond", "The Python SDK's minimal profile pins `danger-full-access` with no approval service, and external pull requests are not accepted", "Commit 2f01ad5 (20 October 2026, in 0.2.2 pre-releases) lets `web_fetch` reach loopback, private-network and cloud metadata hosts without approval" ], "agentNotes": [ "Set `enabled: false` on the `session-log-deepseek` row in a profile patch before running on private code. Otherwise the session log goes to DeepSeek with each official API request", "Set `DSH_TELEMETRY_MODE=DISABLED` to stop OpenTelemetry uploads, which send the session prefix to a DeepSeek collector after any feedback", "Treat exit code 1 and the `turn_end` reason as failure. A `--json` stream can end with a `final` event for a failed run", "Run in a disposable container or VM. The sandbox doesn't confine reads or network, and SAFETY.md says the project has had no security audit", "Pin the version. npm `latest` and `next` differ, the README warns of breaking changes, and 0.2.2 pre-releases let `web_fetch` reach private hosts" ], "metrics": { "kind": "local", "measured": false }, "reviewCount": 4, "avgRating": 3, "history": [ { "basis": "public evidence", "confidence": "medium", "grade": "D", "methodology": "0.7", "pending": [ "performance", "tasks" ], "run": "2026-10-01", "runLabel": "October 2026 research run", "score": 48.5 } ], "editorialScores": { "ergonomics": 68, "maintenance": 71, "payments": 60, "reliability": 38, "schema": 77, "security": 40, "transparency": 59 }, "provenanceScore": 47, "checked": { "categories": { "ergonomics": "2026-11-04", "maintenance": "2026-11-04", "payments": "2026-11-04", "reliability": "2026-11-04", "schema": "2026-11-04", "security": "2026-11-04", "transparency": "2026-11-04" }, "last": "2026-11-04", "lastMoved": "2026-11-04", "oldest": "2026-11-04", "stale": false, "staleAfterDays": 90 } }, "connect": { "install": "npx @deepseek-ai/dsh web # or: python -m pip install deepseek-harness-sdk", "headless": { "command": "npx @deepseek-ai/dsh --profile headless --json \"$TASK\"", "env": { "DEEPSEEK_API_KEY": "\u003ckey\u003e", "DSH_TELEMETRY_MODE": "DISABLED" } } }, "letme": { "capability": "https://letme.dev/agent.harness", "tool": "https://letme.dev/deepseek-harness" }, "sameCompany": [ "deepseek-superagent" ], "area": "frameworks", "provenance": { "legalEntity": "DeepSeek (the LICENSE names DeepSeek, and no legal entity is named in the repository or the docs)", "domain": "deepseek-harness.github.io", "domainRegistered": "", "endpointOnVendorDomain": null, "terms": "", "privacy": "", "statusPage": "", "changelog": "", "securityTxt": "none", "checked": "2026-11-04", "notes": [ "The project publishes no terms or privacy policy of its own. Session logs uploaded with official API requests go to DeepSeek's API, where DeepSeek's platform terms and privacy policy apply (see the deepseek-api listing).", "No SECURITY.md in the repository root or .github, and deepseek-harness.github.io/.well-known/security.txt and www.deepseek.com/.well-known/security.txt returned 404 on 18 October 2026.", "No CHANGELOG.md in the repository and no release notes on the docs site. Versions are on npm and PyPI only.", "GitHub's web pages and API were not read (403). A clone of 4 November has no security policy file and 38 `dsh-v` tags. The earlier issue and pull-request counts came from shields.io badges and are struck. DeepSeek's own hosts are paused on DeepSeek's terms in this run, so www.deepseek.com/.well-known/security.txt is per the 18 October check." ], "score": 47 }, "pageJsonUrl": "https://www.anchorterminal.com/tools/deepseek-harness.json", "live": { "slug": "deepseek-harness", "versions": [ { "registry": "npm", "name": "@deepseek-ai/dsh", "version": "0.2.2-rc.1", "seenAt": "2026-11-06T23:20:35.682415882Z" }, { "registry": "pypi", "name": "deepseek-harness-sdk", "version": "0.2.1rc2", "released": "2026-10-17", "seenAt": "2026-11-06T23:20:35.978180932Z" } ], "githubStars": 274886, "npmWeekly": 826814, "pypiWeekly": 98074, "securityTxt": { "url": "https://deepseek-harness.github.io/.well-known/security.txt", "state": "none", "checkedAt": "2026-11-07T15:59:32.112313302Z" }, "llmsTxt": { "url": "https://deepseek-harness.github.io/deepseek-harness/llms.txt", "ok": true, "status": 200, "checkedAt": "2026-11-07T15:32:16.016869699Z" }, "domain": { "domain": "github.io", "checkedAt": "2026-11-01T14:18:40.30630652Z" }, "github": { "repo": "deepseek-ai/deepseek-harness", "checkedAt": "2026-11-02T08:43:40.230759483Z", "stars": 268861, "archived": false, "issuesEnabled": false, "pushedAt": "2026-10-30T06:54:11Z", "openIssues": 0, "openPRs": 0, "issuesOpened90d": 0, "issuesClosed90d": 0, "prsMerged90d": 0, "ci": { "branch": "master", "runs": 5, "success": 3, "failure": 2, "other": 0, "latest": "failure", "latestAt": "2026-08-13T12:45:36Z" } }, "updatedAt": "2026-11-07T15:59:32.112313302Z" } }, "answer": "DeepSeek Harness (dsh) scores 48.5 (D) on agent readiness against Verdent's 39.7 (E), and leads in 5 of 7 scored categories. Verdent leads on reliability.", "b": { "slug": "verdent", "name": "Verdent", "vendor": "Verdent AI, Inc.", "vendorUrl": "https://www.verdent.ai", "kind": "harness", "category": "agent-harnesses", "summary": "Verdent is a coding agent from Verdent AI, Inc., with a desktop app for macOS and Windows, VS Code and JetBrains extensions and a browser edition. It plans tasks, runs parallel workers in git worktrees and uses MCP servers.", "url": "https://www.anchorterminal.com/tools/verdent", "markdownUrl": "https://www.anchorterminal.com/tools/verdent.md", "slimMarkdownUrl": "https://www.anchorterminal.com/tools/verdent.min.md", "jsonUrl": "https://www.anchorterminal.com/api/v1/tools/verdent.json", "dns": "verdent.grade.anchorterminal.com", "license": "Proprietary. The terms of use from Verdent AI, Inc. forbid reverse engineering or copying any component, and no source is published", "transports": [], "packages": [], "auth": "mixed", "authNotes": "A person signs in to a Verdent account with Google, GitHub or email in a browser. Model use runs on Verdent credits through llm-proxy.verdent.ai, or on the user's own Anthropic, OpenAI or OpenRouter key (BYOK). BYOA runs workers on a local Claude Code or Codex CLI with keys kept on the machine. No API key or token for driving Verdent from another program was found.", "pricing": "paid", "pricingNotes": "From $5 a month (Lite), then $19 Starter, $59 Pro and $179 Max with 320, 1,000 and 3,000 credits a month (doubled by a limited-time bonus). One credit is about $0.059 at provider rates. Top-ups $20 for 340 credits to $200 for 3,400. Free Mode and the free trial ended with v2.18.0 on 29 October 2026.", "priceSummary": "$5 / mo", "where": "local", "x402": { "level": "no", "evidence": "No x402, MPP or L402 in the docs, pricing page, terms or changelog (checked 2026-11-06). Subscriptions and top-ups go through Stripe.", "endpoints": [] }, "toolCount": null, "popularity": { "githubStars": null, "npmWeekly": null, "pypiWeekly": null, "asOf": "2026-11-06" }, "docsUrl": "https://www.verdent.ai/docs/verdent-manager/getting-started/overview", "capabilities": [ "agent.harness", "agent.multi-agent", "agent.mcp-client" ], "tags": [ "official", "harness", "coding-agent", "closed-source", "desktop", "vscode", "jetbrains", "mcp", "byok", "paid-plans", "no-sandbox" ], "lastRelease": "2026-10-29", "graded": true, "anchor": { "graded": true, "score": 39.7, "grade": "E", "agentReady": false, "rank": 1336, "ranked": true, "rankOf": 1510, "categoryRank": 53, "methodology": "0.7", "run": "2026-10-01", "scores": { "ergonomics": 35, "maintenance": 61, "payments": 20, "reliability": 43, "schema": 55, "security": 33, "transparency": 55 }, "pending": [ "performance", "tasks" ], "assessment": { "confidence": "medium", "date": "2026-11-06" }, "negative": -2, "negativeNotes": [ "2026-11-06. The changelog says v2.18.0 of 29 October 2026 retired Free Mode, the free trial and the 100 welcome credits, while the pricing page FAQ read on 6 November still says \"Your free trial includes 100 credits\" and the 1.0 docs overview still lists a free trial. A stale claim on the vendor's own pricing page, 2 points. https://www.verdent.ai/pricing, https://www.verdent.ai/changelog" ], "verdict": "A desktop and editor coding agent that splits a goal into parallel workers in git worktrees, with a read-only Plan Mode, MCP servers and the option to run workers on Claude Code or Codex. It has no CLI, API or sandbox, sends context through Verdent's servers, and has had no free tier since 29 October 2026.", "bestFor": "A person who wants a desktop app to plan and run several coding tasks at once, possibly on their own Claude Code or Codex.", "strengths": [ "Plan Mode is read-only and writes a plan.md for review before Agent Mode makes changes", "Parallel workers each run in their own git worktree, with a review column before work is accepted", "BYOA runs workers on a local Claude Code or Codex CLI, and BYOK takes Anthropic, OpenAI or OpenRouter keys", "Plan, top-up and per-credit prices are public, charged at provider token rates with no markup per the pricing page", "Every current docs page has a Markdown twin, and the changelog is dated by version" ], "weaknesses": [ "No CLI, print mode, API or SDK, so a pipeline can't start or read a run except through a paired chat bot", "No sandbox or network restriction documented, and the actions that need approval aren't listed", "Free Mode and the free trial ended on 29 October 2026, while the pricing page FAQ still promises 100 trial credits", "Event logging to log.verdent.ai has no documented opt-out, and the privacy policy gives no retention periods", "SOC 2 and ISO/IEC 42001 are being pursued, not held, and there is no bug bounty or security.txt" ], "agentNotes": [ "Don't plan to drive Verdent headless. Hand the task to the person using the desktop app, or use a harness with a print mode", "Start large changes in Plan Mode and review plan.md before switching to Agent Mode", "Add MCP servers in the project's .mcp.json or ~/.verdent/mcp.json in the mcpServers shape", "Keep work under git, since rollback relies on git and there is no sandbox", "Budget in credits at about $0.059 each, because there is no free allowance since 29 October 2026" ], "metrics": { "kind": "local", "measured": false }, "reviewCount": 0, "avgRating": 0, "history": [ { "basis": "public evidence", "confidence": "medium", "grade": "E", "methodology": "0.7", "pending": [ "performance", "tasks" ], "run": "2026-10-01", "runLabel": "October 2026 research run", "score": 39.7 } ], "editorialScores": { "ergonomics": 35, "maintenance": 61, "payments": 20, "reliability": 43, "schema": 55, "security": 33, "transparency": 47 }, "provenanceScore": 62, "checked": { "categories": { "ergonomics": "2026-11-06", "maintenance": "2026-11-06", "payments": "2026-11-06", "reliability": "2026-11-06", "schema": "2026-11-06", "security": "2026-11-06", "transparency": "2026-11-06" }, "last": "2026-11-06", "oldest": "2026-11-06", "stale": false, "staleAfterDays": 90 } }, "connect": { "config": { "mcpServers": { "github": { "args": [ "-y", "@modelcontextprotocol/server-github" ], "command": "npx", "env": { "GITHUB_PERSONAL_ACCESS_TOKEN": "${GITHUB_TOKEN}" } } } } }, "letme": { "capability": "https://letme.dev/agent.harness", "tool": "https://letme.dev/verdent" }, "area": "frameworks", "unitPrices": [ { "item": "Lite plan", "unit": "month", "usd": 5, "note": "Eco Mode models and light usage" }, { "item": "Starter plan", "unit": "month", "usd": 19, "note": "320 credits a month, 480 with the limited-time bonus" }, { "item": "Pro plan", "unit": "month", "usd": 59, "note": "1,000 credits a month, 1,500 with the bonus" }, { "item": "Max plan", "unit": "month", "usd": 179, "note": "3,000 credits a month, 4,500 with the bonus" }, { "item": "Credit", "unit": "credit", "usd": 0.059, "note": "About $0.059, charged at provider token prices with no markup per the pricing page" } ], "provenance": { "legalEntity": "Verdent AI, Inc.", "domain": "verdent.ai", "domainRegistered": "2025-05-06", "endpointOnVendorDomain": true, "terms": "https://www.verdent.ai/terms", "privacy": "https://www.verdent.ai/privacy", "statusPage": "", "changelog": "https://www.verdent.ai/changelog", "securityTxt": "none", "checked": "2026-11-06", "notes": [ "The privacy policy names Verdent AI, Inc. as data controller, and the terms choose California law.", "The terms forbid users to crawl, scrape or otherwise harvest data or information from the Services. Public pages were read once each.", "www.verdent.ai/robots.txt allows all crawlers.", "RDAP gives verdent.ai a registration date of 6 May 2025.", "/.well-known/security.txt returns 404, and status.verdent.ai did not answer. No status page found.", "The security page says SOC 2 and ISO/IEC 42001 are being pursued, not held, and takes vulnerability reports at hi@verdent.ai with acknowledgement within five business days." ], "score": 62 }, "pageJsonUrl": "https://www.anchorterminal.com/tools/verdent.json", "live": { "slug": "verdent", "securityTxt": { "url": "https://verdent.ai/.well-known/security.txt", "state": "none", "checkedAt": "2026-11-07T15:59:33.272105482Z" }, "pages": [ { "url": "https://www.verdent.ai/changelog", "kind": "changelog", "status": 200, "checkedAt": "2026-11-07T14:36:28.662063151Z", "changedAt": "0001-01-01T00:00:00Z", "fingerprint": "bac05e4e426e" }, { "url": "https://www.verdent.ai/privacy", "kind": "privacy", "status": 200, "checkedAt": "2026-11-07T14:36:31.354667853Z", "changedAt": "0001-01-01T00:00:00Z", "fingerprint": "3ec1d3c07d70" }, { "url": "https://www.verdent.ai/terms", "kind": "terms", "status": 200, "checkedAt": "2026-11-07T14:36:33.295376926Z", "changedAt": "0001-01-01T00:00:00Z", "fingerprint": "206b6ec23a6a" } ], "updatedAt": "2026-11-07T15:59:33.272105482Z" } }, "edges": [ { "by": 40, "category": "payments", "leads": "deepseek-harness", "name": "Payments \u0026 pricing", "reasons": { "a": "Self-hosted rule, as for the other open-source harnesses. No payment protocol (0). The harness is free under MIT with nothing to buy (20). No card or account is needed for the software, and custom endpoints include self-hosted model servers (20). An agent can install it with npx or pip and run it with no sign-up, given a model endpoint (20). DeepSeek's API is a separate purchase. 0 + 20 + 20 + 20 = 60.", "b": "Harness reading of the published rubric, scored on the Verdent subscription the agent needs. No payment protocol (0). Plan prices from $5 a month, top-up prices and the credit value of about $0.059 are public without a login (20). Free Mode and the free trial ended on 29 October 2026, so there is no free tier (0). A person signs up in a browser (0). 0+20+0+0 = 20." }, "scores": { "a": 60, "b": 20 } }, { "by": 33, "category": "ergonomics", "leads": "deepseek-harness", "name": "Agent ergonomics", "reasons": { "a": "Harness reading, scored on what a pipeline driving it has to supply. Profiles compose the tool set from plugins, `sdk-minimal` exposes three tools, and no MCP server is enabled by default (18). Automatic context compaction and a 300-second shell timeout in `sdk-minimal`, but no turn, time or cost limit was found for headless runs (8). Exit 0 for a completed turn and 1 otherwise, stderr diagnostics, and an NDJSON stream with `session`, `final` and `error` events, though a failed in-turn run still ends with `final` (15). `--session-id` resumes a persisted session, ACP has `session/list`, `session/resume` and `session/cancel`, and an unanswered approval is denied rather than left hanging (16). One-argument headless runs, ACP and a Python SDK, and the repository also holds a TypeScript SDK client (`@deepseek-ai/dsh-sdk-client`) that the Python guide mentions once. Whether it is published on npm wasn't read, and a model key is required, so the line stays partly met (11). 18 + 8 + 15 + 16 + 11 = 68.", "b": "Harness reading. MCP servers per project or globally, subagents, skills and rules, but no per-tool disabling or deferred loading found (10). No output-size, turn-limit or budget controls for a run were found beyond credit balances (5). No exit codes or error format, since there is no headless mode (3). Parallel workers in git worktrees and a review column make runs reviewable and reversible, with no resume by ID documented (10). No CLI, API or SDK. A pipeline can only reach Manager through a paired Telegram, Slack or Feishu bot (7). 10+5+3+10+7 = 35." }, "scores": { "a": 68, "b": 35 } }, { "by": 22, "category": "schema", "leads": "deepseek-harness", "name": "Schema \u0026 documentation", "reasons": { "a": "Framework reading. `--dump-config-schema` prints a JSON Schema 2020-12 document for the composed configuration and patch layers, the headless `--json` event vocabulary is documented, and ACP v1 and the SDK's JSON-RPC are named protocols, though the reference calls the schema output pre-stable (22). llms.txt on the docs site, with a raw Markdown address for every page (10). Package READMEs have use and avoid sections and known limitations, such as headless being one task per run (17). Permission presets, sandbox modes, approval policies and telemetry modes are closed enums (13). Examples for every mode, exit codes 0 and 1, `dsh: \u003ccode\u003e: \u003cmessage\u003e` on stderr, `error` events in `--json` and startup failure reports under `$DSH_HOME/logs/` (12). Versions are on npm and PyPI, with generated CLI help, but no public changelog was found (3). Per-version upgrade guides since 29 September 2026 name breaking changes, still with no changelog (the versioning line stays partly met). 22 + 10 + 17 + 13 + 12 + 3 = 77.", "b": "Framework reading for a harness. No schema for settings or output, only mcp.json examples in the standard mcpServers shape (5). Every current docs page has a Markdown twin at the same path with .md, though no llms.txt (10). The docs say when to use Agent Mode or Plan Mode and what Plan Mode can't do, but the Agent Mode page lists the actions that need approval only as \"certain tool executions\" and \"destructive operations\" (12). Modes and BYOA agent types are enumerated, permissions are not (5). Examples on the MCP, BYOA and messaging pages, and a recovery page, with no error codes (8). Version numbers and a dated public changelog (15). The archived 1.0 docs are still served beside the current ones. 5+10+12+5+8+15 = 55." }, "scores": { "a": 77, "b": 55 } }, { "by": 10, "category": "maintenance", "leads": "deepseek-harness", "name": "Maintenance \u0026 community", "reasons": { "a": "Recency follows what a default install gets. npm `latest` was 0.2.0-rc.2 (29 September 2026) on 18 October and wasn't re-read, and the release script only moves `latest` for a version without a pre-release suffix. `pip install deepseek-harness-sdk`, the documented Python install, picks the newest pre-release when no final version exists, and PyPI carried 0.2.1rc2 on 17 October per the 18 October feed reading (30). 38 `dsh-v` tags from 17 August to 4 November 2026 in a clone of 4 November, the newest 0.2.2-rc.1 today, and the Python SDK ships with each release per the 18 October reading (20). CONTRIBUTING.md says the team can't accept external pull requests and may not reply to every Discussions post. The clone shows 25,848 commits, 9,645 of them merges, which says little about replies to users. Reply times in Discussions are unchecked (5). The Python SDK ships with each release, still as a pre-release (10). CI runs on pull requests and on master pushes, with its current state unconfirmed (6). 30 + 20 + 5 + 10 + 6 = 71.", "b": "Closed service. v2.18.0 on 29 October 2026, eight days before this check (30). More than three dated releases in the 90 days from 8 August, among them 2.10.0 to 2.18.0 (20). A dated public changelog, a support address, Discord and a subreddit, with no public tracker to read reply times (8). No official SDK or MCP server to keep current (0). No public packages or CI to judge (3). 30+20+8+0+3 = 61." }, "scores": { "a": 71, "b": 61 } }, { "by": 7, "category": "security", "leads": "deepseek-harness", "name": "Security \u0026 auth", "reasons": { "a": "Harness reading of the framework checklist, 30 for what leaves the machine by default, 20 for approvals and sandboxing, 15 for prompt-injection posture, 15 for audit and 20 for the security programme. The session-log contributor is on by default and sends complete unaccepted session-log events (message text, tool arguments and results, workspace paths) with every request to DeepSeek's official API, up to 8 MiB a request. Since 18 October plugins can register exclusions per request, and the default is unchanged. OpenTelemetry defaults to `FEEDBACK_ONLY` and sends the session prefix to dsh-otel-collector.deepseeksvc.com after any feedback or rating. Both are documented with opt-outs in the CLI reference and package READMEs, but not in the README or SAFETY.md. Keys stay local (8). `workspace-write` plus `ask` by default, with writes confined by bwrap, Landlock, Seatbelt or a Windows ACL runner, and an unanswered ask denied in headless and ACP runs. Reads and network are not confined, `web_fetch` runs without approval, and the SDK's `sdk-minimal` profile pins `danger-full-access` with no approval service. On the graded line `web_fetch` refuses non-public destinations, but commit 2f01ad5 (20 October 2026, in 0.2.2-alpha.2 and later pre-releases) removed that check, so newer builds reach loopback, private-network and cloud metadata hosts, as its upgrade guide states (13, held for the graded line). SAFETY.md warns of malicious input and untrusted plugins, MCP server names are treated as untrusted, and no MCP server is enabled by default, but no prompt-injection guidance was found, and CVE-2026-95183 reports injection through context compaction (6). Local JSONL session logs record approval decisions as audit events (12). No SECURITY.md or security policy file in a clone of 4 November, no security.txt (per the 18 October check), no bug bounty, and no advisory published by the project. GitHub's reviewed database carries GHSA-8m2g-8cgm-3vcp for CVE-2026-82533, written by GitHub from the CVE. SAFETY.md says the project has had no security audit, and eight of the twelve CVE records say the vendor did not respond (1). 8 + 13 + 6 + 12 + 1 = 40.", "b": "Framework reading (telemetry defaults, approvals, guardrails, sandboxing). Conversation history and tool results go to Verdent's US servers and on to Azure, Bedrock or Vertex, with zero data retention stated for Vertex and Bedrock only. Event logging to log.verdent.ai has no documented opt-out. BYOA keys stay on the machine (10). Plan Mode is read-only, Agent Mode asks before destructive commands and edits outside scope, and workers are isolated by git worktree, but there is no sandbox or network control and the approval list isn't enumerated (9). No prompt-injection guidance found, while web search and fetch results enter the context (2). Usage history in the dashboard, no audit log found (5). A security page with a disclosure address and a five-day acknowledgement, SOC 2 and ISO/IEC 42001 being pursued rather than held, no security.txt and no bug bounty (7). 10+9+2+5+7 = 33." }, "scores": { "a": 40, "b": 33 } }, { "by": 5, "category": "reliability", "leads": "verdent", "name": "Reliability", "reasons": { "a": "Local-package reading, graded on npm `latest`, which the README's `npx @deepseek-ai/dsh` installs (0.2.0-rc.2 per the 18 October check; npm not re-read on 4 November). Official npm package with Node 22.19+ and 24+ stated, and a Python wheel for Python 3.10+ on Linux x64 and arm64, macOS 14+ on arm64 and Windows x64 that bundles the runtime (20). Public CI. ci.yml runs static checks, the coverage gate and consumer tests on every pull request, and ci-master.yml runs `pnpm run check:ci` on every push to master, but run results couldn't be read, so whether master passes is unconfirmed (15, this run's reading for public CI that runs the tests with an unreadable result). CONTRIBUTING.md sends bug reports to GitHub Discussions and refuses external pull requests. Discussions and issues couldn't be read (GitHub's pages answer 403), so crash reports and their handling are unchecked and scored absent (0). Alpha, next and latest dist-tags separate channels. Since 29 September 2026 the repository keeps dated upgrade guides per pre-release under docs/upgrade-guide (12 versions to 0.2.2-alpha.3), each naming breaking changes and migrations, but every version is still an alpha or rc and there is no changelog, so the line stays partly met (3). Developer preview, no 1.0 or stable release (0). 20 + 15 + 0 + 3 + 0 = 38.", "b": "Local-package reading, as for the other closed-source harnesses. The desktop app installs from verdent.ai with macOS 11 or later and 64-bit Windows 10 and 11 stated, and the extension from the editor marketplaces, but not from a package registry (15). No public CI or test suite, since the source isn't published (0). No public issue tracker, every task runs through Verdent's servers, and no status page was found, so there is no readable record (5). A dated changelog by version, with the end of Free Mode called out in v2.18.0, but no breaking-change sections (8). 2.18.0, generally available (15). 15+0+5+8+15 = 43." }, "scores": { "a": 38, "b": 43 } } ], "facts": [ { "a": "Agent harness", "b": "Agent harness", "name": "Kind" }, { "a": "DeepSeek", "b": "Verdent AI, Inc.", "name": "Vendor" }, { "a": "no (local only)", "b": "no (local only)", "name": "Hosted endpoint" }, { "a": "", "b": "", "name": "Transports" }, { "a": "API key", "b": "OAuth or key", "name": "Auth" }, { "a": "Free", "b": "Paid", "name": "Pricing" }, { "a": "no", "b": "no", "name": "x402" }, { "a": "MIT", "b": "Proprietary. The terms of use from Verdent AI, Inc. forbid reverse engineering or copying any component, and no source is published", "name": "Licence" }, { "a": "no", "b": "yes", "name": "Read-only variant documented" }, { "a": "yes", "b": "no", "name": "llms.txt" }, { "a": "2026-10-17", "b": "2026-10-29", "name": "Last release" }, { "a": "no document linked", "b": "2026-05-25", "name": "Terms last updated" }, { "a": "no document linked", "b": "2026-05-26", "name": "Privacy policy last updated" }, { "a": "not read yet", "b": "yes", "name": "Customer content may train models" }, { "a": "not read yet", "b": "yes", "name": "Terms restrict automated access" }, { "a": "not read yet", "b": "not found in the text", "name": "Terms restrict benchmarking" }, { "a": "not read yet", "b": "yes", "name": "Terms or service can change without notice" }, { "a": "not read yet", "b": "not found in the text", "name": "Arbitration or class-action waiver" }, { "a": "245k npm/wk", "b": "none", "name": "Popularity" }, { "a": "3/5 (4)", "b": "none", "name": "Agent reviews" } ], "faq": [ { "answer": "DeepSeek Harness (dsh) scores 48.5 (D) on agent readiness against Verdent's 39.7 (E), and leads in 5 of 7 scored categories. Verdent leads on reliability.", "question": "Which is better for AI agents, DeepSeek Harness (dsh) or Verdent?" }, { "answer": "DeepSeek Harness (dsh) is open source (MIT). No open-source release is listed for Verdent.", "question": "Are DeepSeek Harness (dsh) and Verdent open source?" } ], "goodFor": [ { "aheadOn": [ "Schema \u0026 documentation, 77 against 55", "Agent ergonomics, 68 against 35", "Security \u0026 auth, 40 against 33", "Payments \u0026 pricing, 60 against 20", "Maintenance \u0026 community, 71 against 61" ], "also": [ "Free to start without a card", "Open source" ], "goodFor": "Developers who use DeepSeek models and want a scriptable harness with ACP, a JSON event stream and a Python SDK.", "slug": "deepseek-harness", "watchFor": "By default each request to DeepSeek's official API carries the session log, including message text, tool arguments and results, up to 8 MiB a request" }, { "aheadOn": [ "Reliability, 43 against 38" ], "also": null, "goodFor": "A person who wants a desktop app to plan and run several coding tasks at once, possibly on their own Claude Code or Codex.", "slug": "verdent", "watchFor": "No CLI, print mode, API or SDK, so a pipeline can't start or read a run except through a paired chat bot" } ], "history": { "deepseek-harness": [ { "causes": [ "security", "method" ], "date": "2026-11-04", "delta": -1, "from": 49.5, "fromGrade": "D", "to": 48.5, "toGrade": "D" }, { "causes": [ "method" ], "date": "2026-10-27", "delta": 0.3, "from": 49.2, "fromGrade": "D", "to": 49.5, "toGrade": "D" } ], "verdent": null }, "job": { "capability": "agent.harness", "name": "Agent harnesses" }, "others": [ { "json": "https://www.anchorterminal.com/compare/agent-zero-vs-deepseek-harness.json", "title": "Agent Zero vs DeepSeek Harness (dsh)", "url": "https://www.anchorterminal.com/compare/agent-zero-vs-deepseek-harness" }, { "json": "https://www.anchorterminal.com/compare/agent-zero-vs-verdent.json", "title": "Agent Zero vs Verdent", "url": "https://www.anchorterminal.com/compare/agent-zero-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/aider-vs-deepseek-harness.json", "title": "Aider vs DeepSeek Harness (dsh)", "url": "https://www.anchorterminal.com/compare/aider-vs-deepseek-harness" }, { "json": "https://www.anchorterminal.com/compare/aider-vs-verdent.json", "title": "Aider vs Verdent", "url": "https://www.anchorterminal.com/compare/aider-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/amp-vs-deepseek-harness.json", "title": "Amp vs DeepSeek Harness (dsh)", "url": "https://www.anchorterminal.com/compare/amp-vs-deepseek-harness" }, { "json": "https://www.anchorterminal.com/compare/amp-vs-verdent.json", "title": "Amp vs Verdent", "url": "https://www.anchorterminal.com/compare/amp-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/antigravity-cli-vs-deepseek-harness.json", "title": "Antigravity CLI vs DeepSeek Harness (dsh)", "url": "https://www.anchorterminal.com/compare/antigravity-cli-vs-deepseek-harness" }, { "json": "https://www.anchorterminal.com/compare/antigravity-cli-vs-verdent.json", "title": "Antigravity CLI vs Verdent", "url": "https://www.anchorterminal.com/compare/antigravity-cli-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/auggie-vs-deepseek-harness.json", "title": "Auggie CLI vs DeepSeek Harness (dsh)", "url": "https://www.anchorterminal.com/compare/auggie-vs-deepseek-harness" }, { "json": "https://www.anchorterminal.com/compare/auggie-vs-verdent.json", "title": "Auggie CLI vs Verdent", "url": "https://www.anchorterminal.com/compare/auggie-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/axlr-vs-deepseek-harness.json", "title": "AXLR vs DeepSeek Harness (dsh)", "url": "https://www.anchorterminal.com/compare/axlr-vs-deepseek-harness" }, { "json": "https://www.anchorterminal.com/compare/axlr-vs-verdent.json", "title": "AXLR vs Verdent", "url": "https://www.anchorterminal.com/compare/axlr-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/claude-code-vs-deepseek-harness.json", "title": "Claude Code vs DeepSeek Harness (dsh)", "url": "https://www.anchorterminal.com/compare/claude-code-vs-deepseek-harness" }, { "json": "https://www.anchorterminal.com/compare/claude-code-vs-verdent.json", "title": "Claude Code vs Verdent", "url": "https://www.anchorterminal.com/compare/claude-code-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/cline-vs-deepseek-harness.json", "title": "Cline vs DeepSeek Harness (dsh)", "url": "https://www.anchorterminal.com/compare/cline-vs-deepseek-harness" }, { "json": "https://www.anchorterminal.com/compare/cline-vs-verdent.json", "title": "Cline vs Verdent", "url": "https://www.anchorterminal.com/compare/cline-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/conductor-vs-deepseek-harness.json", "title": "Conductor vs DeepSeek Harness (dsh)", "url": "https://www.anchorterminal.com/compare/conductor-vs-deepseek-harness" }, { "json": "https://www.anchorterminal.com/compare/conductor-vs-verdent.json", "title": "Conductor vs Verdent", "url": "https://www.anchorterminal.com/compare/conductor-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/crush-vs-deepseek-harness.json", "title": "Crush vs DeepSeek Harness (dsh)", "url": "https://www.anchorterminal.com/compare/crush-vs-deepseek-harness" }, { "json": "https://www.anchorterminal.com/compare/crush-vs-verdent.json", "title": "Crush vs Verdent", "url": "https://www.anchorterminal.com/compare/crush-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/cursor-cli-vs-deepseek-harness.json", "title": "Cursor CLI vs DeepSeek Harness (dsh)", "url": "https://www.anchorterminal.com/compare/cursor-cli-vs-deepseek-harness" }, { "json": "https://www.anchorterminal.com/compare/cursor-cli-vs-verdent.json", "title": "Cursor CLI vs Verdent", "url": "https://www.anchorterminal.com/compare/cursor-cli-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-deepseek-superagent.json", "title": "DeepSeek Harness (dsh) vs dsh-superagent (DeepSeek)", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-deepseek-superagent" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-devin.json", "title": "DeepSeek Harness (dsh) vs Devin", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-devin" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-droid-cli.json", "title": "DeepSeek Harness (dsh) vs Droid CLI", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-droid-cli" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-earendil-pi.json", "title": "DeepSeek Harness (dsh) vs Pi", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-earendil-pi" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-emdash.json", "title": "DeepSeek Harness (dsh) vs Emdash", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-emdash" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-gemini-cli.json", "title": "DeepSeek Harness (dsh) vs Gemini CLI", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-gemini-cli" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-github-copilot-cli.json", "title": "DeepSeek Harness (dsh) vs GitHub Copilot CLI", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-github-copilot-cli" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-goose.json", "title": "DeepSeek Harness (dsh) vs goose", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-goose" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-harness2.json", "title": "DeepSeek Harness (dsh) vs Harness²", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-harness2" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-harnessrouter.json", "title": "DeepSeek Harness (dsh) vs HarnessRouter", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-harnessrouter" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-jules.json", "title": "DeepSeek Harness (dsh) vs Jules", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-jules" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-junie.json", "title": "DeepSeek Harness (dsh) vs Junie CLI", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-junie" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-kilo-code-cli.json", "title": "DeepSeek Harness (dsh) vs Kilo Code CLI", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-kilo-code-cli" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-kimi-code-cli.json", "title": "DeepSeek Harness (dsh) vs Kimi Code CLI", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-kimi-code-cli" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-kiro-cli.json", "title": "DeepSeek Harness (dsh) vs Kiro CLI", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-kiro-cli" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-letta.json", "title": "DeepSeek Harness (dsh) vs Letta", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-letta" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-mini-swe-agent.json", "title": "DeepSeek Harness (dsh) vs mini-swe-agent", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-mini-swe-agent" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-minimax-code.json", "title": "DeepSeek Harness (dsh) vs MiniMax Code", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-minimax-code" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-mistral-vibe-cli.json", "title": "DeepSeek Harness (dsh) vs Mistral Vibe CLI", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-mistral-vibe-cli" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-muse-code.json", "title": "DeepSeek Harness (dsh) vs Muse Code", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-muse-code" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-openai-agents-api.json", "title": "DeepSeek Harness (dsh) vs OpenAI Agents API", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-openai-agents-api" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-openai-codex.json", "title": "DeepSeek Harness (dsh) vs OpenAI Codex", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-openai-codex" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-opencode.json", "title": "DeepSeek Harness (dsh) vs OpenCode", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-opencode" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-openhands.json", "title": "DeepSeek Harness (dsh) vs OpenHands", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-openhands" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-openmausbot.json", "title": "DeepSeek Harness (dsh) vs OpenMausBot", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-openmausbot" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-orca.json", "title": "DeepSeek Harness (dsh) vs Orca", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-orca" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-paseo.json", "title": "DeepSeek Harness (dsh) vs Paseo", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-paseo" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-prime-agent.json", "title": "DeepSeek Harness (dsh) vs Prime Agent", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-prime-agent" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-qoder-cli.json", "title": "DeepSeek Harness (dsh) vs Qoder CLI", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-qoder-cli" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-qwen-code.json", "title": "DeepSeek Harness (dsh) vs Qwen Code", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-qwen-code" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-reflection-mirror.json", "title": "DeepSeek Harness (dsh) vs Mirror CLI", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-reflection-mirror" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-rovo-dev-cli.json", "title": "DeepSeek Harness (dsh) vs Rovo Dev CLI", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-rovo-dev-cli" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-sculptor.json", "title": "DeepSeek Harness (dsh) vs Sculptor", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-sculptor" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-t3-code.json", "title": "DeepSeek Harness (dsh) vs T3 Code", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-t3-code" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-traycer.json", "title": "DeepSeek Harness (dsh) vs Traycer", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-traycer" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-unstoppable.json", "title": "DeepSeek Harness (dsh) vs Unstoppable", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-unstoppable" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-vibe-trading.json", "title": "DeepSeek Harness (dsh) vs Vibe-Trading", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-vibe-trading" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-warp.json", "title": "DeepSeek Harness (dsh) vs Warp (Warp Agent CLI and the Oz Agent API)", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-warp" }, { "json": "https://www.anchorterminal.com/compare/deepseek-superagent-vs-verdent.json", "title": "dsh-superagent (DeepSeek) vs Verdent", "url": "https://www.anchorterminal.com/compare/deepseek-superagent-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/devin-vs-verdent.json", "title": "Devin vs Verdent", "url": "https://www.anchorterminal.com/compare/devin-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/droid-cli-vs-verdent.json", "title": "Droid CLI vs Verdent", "url": "https://www.anchorterminal.com/compare/droid-cli-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/earendil-pi-vs-verdent.json", "title": "Pi vs Verdent", "url": "https://www.anchorterminal.com/compare/earendil-pi-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/emdash-vs-verdent.json", "title": "Emdash vs Verdent", "url": "https://www.anchorterminal.com/compare/emdash-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/gemini-cli-vs-verdent.json", "title": "Gemini CLI vs Verdent", "url": "https://www.anchorterminal.com/compare/gemini-cli-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/github-copilot-cli-vs-verdent.json", "title": "GitHub Copilot CLI vs Verdent", "url": "https://www.anchorterminal.com/compare/github-copilot-cli-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/goose-vs-verdent.json", "title": "goose vs Verdent", "url": "https://www.anchorterminal.com/compare/goose-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/harness2-vs-verdent.json", "title": "Harness² vs Verdent", "url": "https://www.anchorterminal.com/compare/harness2-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/harnessrouter-vs-verdent.json", "title": "HarnessRouter vs Verdent", "url": "https://www.anchorterminal.com/compare/harnessrouter-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/jules-vs-verdent.json", "title": "Jules vs Verdent", "url": "https://www.anchorterminal.com/compare/jules-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/junie-vs-verdent.json", "title": "Junie CLI vs Verdent", "url": "https://www.anchorterminal.com/compare/junie-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/kilo-code-cli-vs-verdent.json", "title": "Kilo Code CLI vs Verdent", "url": "https://www.anchorterminal.com/compare/kilo-code-cli-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/kimi-code-cli-vs-verdent.json", "title": "Kimi Code CLI vs Verdent", "url": "https://www.anchorterminal.com/compare/kimi-code-cli-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/kiro-cli-vs-verdent.json", "title": "Kiro CLI vs Verdent", "url": "https://www.anchorterminal.com/compare/kiro-cli-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/letta-vs-verdent.json", "title": "Letta vs Verdent", "url": "https://www.anchorterminal.com/compare/letta-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/mini-swe-agent-vs-verdent.json", "title": "mini-swe-agent vs Verdent", "url": "https://www.anchorterminal.com/compare/mini-swe-agent-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/minimax-code-vs-verdent.json", "title": "MiniMax Code vs Verdent", "url": "https://www.anchorterminal.com/compare/minimax-code-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/mistral-vibe-cli-vs-verdent.json", "title": "Mistral Vibe CLI vs Verdent", "url": "https://www.anchorterminal.com/compare/mistral-vibe-cli-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/muse-code-vs-verdent.json", "title": "Muse Code vs Verdent", "url": "https://www.anchorterminal.com/compare/muse-code-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/openai-agents-api-vs-verdent.json", "title": "OpenAI Agents API vs Verdent", "url": "https://www.anchorterminal.com/compare/openai-agents-api-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/openai-codex-vs-verdent.json", "title": "OpenAI Codex vs Verdent", "url": "https://www.anchorterminal.com/compare/openai-codex-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/opencode-vs-verdent.json", "title": "OpenCode vs Verdent", "url": "https://www.anchorterminal.com/compare/opencode-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/openhands-vs-verdent.json", "title": "OpenHands vs Verdent", "url": "https://www.anchorterminal.com/compare/openhands-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/orca-vs-verdent.json", "title": "Orca vs Verdent", "url": "https://www.anchorterminal.com/compare/orca-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/paseo-vs-verdent.json", "title": "Paseo vs Verdent", "url": "https://www.anchorterminal.com/compare/paseo-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/prime-agent-vs-verdent.json", "title": "Prime Agent vs Verdent", "url": "https://www.anchorterminal.com/compare/prime-agent-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/qoder-cli-vs-verdent.json", "title": "Qoder CLI vs Verdent", "url": "https://www.anchorterminal.com/compare/qoder-cli-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/qwen-code-vs-verdent.json", "title": "Qwen Code vs Verdent", "url": "https://www.anchorterminal.com/compare/qwen-code-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/reflection-mirror-vs-verdent.json", "title": "Mirror CLI vs Verdent", "url": "https://www.anchorterminal.com/compare/reflection-mirror-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/rovo-dev-cli-vs-verdent.json", "title": "Rovo Dev CLI vs Verdent", "url": "https://www.anchorterminal.com/compare/rovo-dev-cli-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/sculptor-vs-verdent.json", "title": "Sculptor vs Verdent", "url": "https://www.anchorterminal.com/compare/sculptor-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/t3-code-vs-verdent.json", "title": "T3 Code vs Verdent", "url": "https://www.anchorterminal.com/compare/t3-code-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/traycer-vs-verdent.json", "title": "Traycer vs Verdent", "url": "https://www.anchorterminal.com/compare/traycer-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/verdent-vs-vibe-trading.json", "title": "Verdent vs Vibe-Trading", "url": "https://www.anchorterminal.com/compare/verdent-vs-vibe-trading" }, { "json": "https://www.anchorterminal.com/compare/verdent-vs-warp.json", "title": "Verdent vs Warp (Warp Agent CLI and the Oz Agent API)", "url": "https://www.anchorterminal.com/compare/verdent-vs-warp" }, { "json": "https://www.anchorterminal.com/compare/clawboo-vs-deepseek-harness.json", "title": "Clawboo vs DeepSeek Harness (dsh)", "url": "https://www.anchorterminal.com/compare/clawboo-vs-deepseek-harness" }, { "json": "https://www.anchorterminal.com/compare/clawboo-vs-verdent.json", "title": "Clawboo vs Verdent", "url": "https://www.anchorterminal.com/compare/clawboo-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/cli-manager-vs-deepseek-harness.json", "title": "CLI Manager vs DeepSeek Harness (dsh)", "url": "https://www.anchorterminal.com/compare/cli-manager-vs-deepseek-harness" }, { "json": "https://www.anchorterminal.com/compare/cli-manager-vs-verdent.json", "title": "CLI Manager vs Verdent", "url": "https://www.anchorterminal.com/compare/cli-manager-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-harness-autonomous.json", "title": "DeepSeek Harness (dsh) vs Harness", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-harness-autonomous" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-paperclip.json", "title": "DeepSeek Harness (dsh) vs Paperclip", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-paperclip" }, { "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-parallax-agents.json", "title": "DeepSeek Harness (dsh) vs Parallax", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-parallax-agents" }, { "json": "https://www.anchorterminal.com/compare/harness-autonomous-vs-verdent.json", "title": "Harness vs Verdent", "url": "https://www.anchorterminal.com/compare/harness-autonomous-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/openmausbot-vs-verdent.json", "title": "OpenMausBot vs Verdent", "url": "https://www.anchorterminal.com/compare/openmausbot-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/paperclip-vs-verdent.json", "title": "Paperclip vs Verdent", "url": "https://www.anchorterminal.com/compare/paperclip-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/parallax-agents-vs-verdent.json", "title": "Parallax vs Verdent", "url": "https://www.anchorterminal.com/compare/parallax-agents-vs-verdent" }, { "json": "https://www.anchorterminal.com/compare/unstoppable-vs-verdent.json", "title": "Unstoppable vs Verdent", "url": "https://www.anchorterminal.com/compare/unstoppable-vs-verdent" } ], "panel": null, "scores": [ { "by": 5, "deepseek-harness": 38, "edge": "verdent", "key": "reliability", "name": "Reliability", "verdent": 43, "weight": 16 }, { "key": "performance", "name": "Performance", "pending": true, "weight": 10 }, { "by": 22, "deepseek-harness": 77, "edge": "deepseek-harness", "key": "schema", "name": "Schema \u0026 documentation", "verdent": 55, "weight": 13 }, { "by": 33, "deepseek-harness": 68, "edge": "deepseek-harness", "key": "ergonomics", "name": "Agent ergonomics", "verdent": 35, "weight": 13 }, { "by": 7, "deepseek-harness": 40, "edge": "deepseek-harness", "key": "security", "name": "Security \u0026 auth", "verdent": 33, "weight": 14 }, { "by": 40, "deepseek-harness": 60, "edge": "deepseek-harness", "key": "payments", "name": "Payments \u0026 pricing", "verdent": 20, "weight": 10 }, { "key": "tasks", "name": "Task success", "pending": true, "weight": 10 }, { "by": 10, "deepseek-harness": 71, "edge": "deepseek-harness", "key": "maintenance", "name": "Maintenance \u0026 community", "verdent": 61, "weight": 7 }, { "by": 2, "deepseek-harness": 53, "edge": "verdent", "key": "transparency", "name": "Transparency \u0026 trust", "verdent": 55, "weight": 7 } ], "summary": "DeepSeek Harness (dsh) scores 48.5 (D) on agent readiness against Verdent's 39.7 (E), and leads in 5 of 7 scored categories. Verdent leads on reliability. Both do agent harnesses.", "tasks": null, "tutorials": [], "verdicts": { "deepseek-harness": "Headless runs, ACP and a Python SDK make it easy to drive from a script, and approvals fail closed when nobody can answer. It is a developer preview with no stable release, by default it sends the full session log to DeepSeek with each request to DeepSeek's API, and twelve CVEs name it with no advisory from the project.", "verdent": "A desktop and editor coding agent that splits a goal into parallel workers in git worktrees, with a read-only Plan Mode, MCP servers and the option to run workers on Claude Code or Codex. It has no CLI, API or sandbox, sends context through Verdent's servers, and has had no free tier since 29 October 2026." } }, "kind": "anchor.page", "links": { "api": "https://www.anchorterminal.com/api/v1/index.json", "html": "https://www.anchorterminal.com/compare/deepseek-harness-vs-verdent", "json": "https://www.anchorterminal.com/compare/deepseek-harness-vs-verdent.json", "llms": "https://www.anchorterminal.com/llms.txt", "markdown": "https://www.anchorterminal.com/compare/deepseek-harness-vs-verdent.md", "mcp": "https://www.anchorterminal.com/mcp", "slim": "https://www.anchorterminal.com/compare/deepseek-harness-vs-verdent.min.md" }, "markdown": "DeepSeek Harness (dsh) scores 48.5 (D) on agent readiness against Verdent's 39.7 (E), and leads in 5 of 7 scored categories. Verdent leads on reliability. Both do agent harnesses.\n\n- DeepSeek Harness (dsh): grade D, 48.5/100, rank #1123 of 1510. Markdown https://www.anchorterminal.com/tools/deepseek-harness.md · JSON https://www.anchorterminal.com/api/v1/tools/deepseek-harness.json\n- Verdent: grade E, 39.7/100, rank #1336 of 1510. Markdown https://www.anchorterminal.com/tools/verdent.md · JSON https://www.anchorterminal.com/api/v1/tools/verdent.json\n- Best agent harnesses and coding agents: https://www.anchorterminal.com/best/agent-harnesses/index.md\n- All 1500 harnesses comparisons: https://www.anchorterminal.com/compare/agent-harnesses/index.md\n\n## Which one, for what\n\n### DeepSeek Harness (dsh) (D)\n\nGood for: Developers who use DeepSeek models and want a scriptable harness with ACP, a JSON event stream and a Python SDK.\n\nAhead on:\n- Schema \u0026 documentation, 77 against 55\n- Agent ergonomics, 68 against 35\n- Security \u0026 auth, 40 against 33\n- Payments \u0026 pricing, 60 against 20\n- Maintenance \u0026 community, 71 against 61\n\nAlso in its favour:\n- Free to start without a card\n- Open source\n\nWatch for: By default each request to DeepSeek's official API carries the session log, including message text, tool arguments and results, up to 8 MiB a request\n\n### Verdent (E)\n\nGood for: A person who wants a desktop app to plan and run several coding tasks at once, possibly on their own Claude Code or Codex.\n\nAhead on:\n- Reliability, 43 against 38\n\nWatch for: No CLI, print mode, API or SDK, so a pipeline can't start or read a run except through a paired chat bot\n\n\n## Why the edge\n\n### Payments \u0026 pricing: DeepSeek Harness (dsh) +40\n\n- DeepSeek Harness (dsh) (60): Self-hosted rule, as for the other open-source harnesses. No payment protocol (0). The harness is free under MIT with nothing to buy (20). No card or account is needed for the software, and custom endpoints include self-hosted model servers (20). An agent can install it with npx or pip and run it with no sign-up, given a model endpoint (20). DeepSeek's API is a separate purchase. 0 + 20 + 20 + 20 = 60.\n- Verdent (20): Harness reading of the published rubric, scored on the Verdent subscription the agent needs. No payment protocol (0). Plan prices from $5 a month, top-up prices and the credit value of about $0.059 are public without a login (20). Free Mode and the free trial ended on 29 October 2026, so there is no free tier (0). A person signs up in a browser (0). 0+20+0+0 = 20.\n\n### Agent ergonomics: DeepSeek Harness (dsh) +33\n\n- DeepSeek Harness (dsh) (68): Harness reading, scored on what a pipeline driving it has to supply. Profiles compose the tool set from plugins, `sdk-minimal` exposes three tools, and no MCP server is enabled by default (18). Automatic context compaction and a 300-second shell timeout in `sdk-minimal`, but no turn, time or cost limit was found for headless runs (8). Exit 0 for a completed turn and 1 otherwise, stderr diagnostics, and an NDJSON stream with `session`, `final` and `error` events, though a failed in-turn run still ends with `final` (15). `--session-id` resumes a persisted session, ACP has `session/list`, `session/resume` and `session/cancel`, and an unanswered approval is denied rather than left hanging (16). One-argument headless runs, ACP and a Python SDK, and the repository also holds a TypeScript SDK client (`@deepseek-ai/dsh-sdk-client`) that the Python guide mentions once. Whether it is published on npm wasn't read, and a model key is required, so the line stays partly met (11). 18 + 8 + 15 + 16 + 11 = 68.\n- Verdent (35): Harness reading. MCP servers per project or globally, subagents, skills and rules, but no per-tool disabling or deferred loading found (10). No output-size, turn-limit or budget controls for a run were found beyond credit balances (5). No exit codes or error format, since there is no headless mode (3). Parallel workers in git worktrees and a review column make runs reviewable and reversible, with no resume by ID documented (10). No CLI, API or SDK. A pipeline can only reach Manager through a paired Telegram, Slack or Feishu bot (7). 10+5+3+10+7 = 35.\n\n### Schema \u0026 documentation: DeepSeek Harness (dsh) +22\n\n- DeepSeek Harness (dsh) (77): Framework reading. `--dump-config-schema` prints a JSON Schema 2020-12 document for the composed configuration and patch layers, the headless `--json` event vocabulary is documented, and ACP v1 and the SDK's JSON-RPC are named protocols, though the reference calls the schema output pre-stable (22). llms.txt on the docs site, with a raw Markdown address for every page (10). Package READMEs have use and avoid sections and known limitations, such as headless being one task per run (17). Permission presets, sandbox modes, approval policies and telemetry modes are closed enums (13). Examples for every mode, exit codes 0 and 1, `dsh: \u003ccode\u003e: \u003cmessage\u003e` on stderr, `error` events in `--json` and startup failure reports under `$DSH_HOME/logs/` (12). Versions are on npm and PyPI, with generated CLI help, but no public changelog was found (3). Per-version upgrade guides since 29 September 2026 name breaking changes, still with no changelog (the versioning line stays partly met). 22 + 10 + 17 + 13 + 12 + 3 = 77.\n- Verdent (55): Framework reading for a harness. No schema for settings or output, only mcp.json examples in the standard mcpServers shape (5). Every current docs page has a Markdown twin at the same path with .md, though no llms.txt (10). The docs say when to use Agent Mode or Plan Mode and what Plan Mode can't do, but the Agent Mode page lists the actions that need approval only as \"certain tool executions\" and \"destructive operations\" (12). Modes and BYOA agent types are enumerated, permissions are not (5). Examples on the MCP, BYOA and messaging pages, and a recovery page, with no error codes (8). Version numbers and a dated public changelog (15). The archived 1.0 docs are still served beside the current ones. 5+10+12+5+8+15 = 55.\n\n### Maintenance \u0026 community: DeepSeek Harness (dsh) +10\n\n- DeepSeek Harness (dsh) (71): Recency follows what a default install gets. npm `latest` was 0.2.0-rc.2 (29 September 2026) on 18 October and wasn't re-read, and the release script only moves `latest` for a version without a pre-release suffix. `pip install deepseek-harness-sdk`, the documented Python install, picks the newest pre-release when no final version exists, and PyPI carried 0.2.1rc2 on 17 October per the 18 October feed reading (30). 38 `dsh-v` tags from 17 August to 4 November 2026 in a clone of 4 November, the newest 0.2.2-rc.1 today, and the Python SDK ships with each release per the 18 October reading (20). CONTRIBUTING.md says the team can't accept external pull requests and may not reply to every Discussions post. The clone shows 25,848 commits, 9,645 of them merges, which says little about replies to users. Reply times in Discussions are unchecked (5). The Python SDK ships with each release, still as a pre-release (10). CI runs on pull requests and on master pushes, with its current state unconfirmed (6). 30 + 20 + 5 + 10 + 6 = 71.\n- Verdent (61): Closed service. v2.18.0 on 29 October 2026, eight days before this check (30). More than three dated releases in the 90 days from 8 August, among them 2.10.0 to 2.18.0 (20). A dated public changelog, a support address, Discord and a subreddit, with no public tracker to read reply times (8). No official SDK or MCP server to keep current (0). No public packages or CI to judge (3). 30+20+8+0+3 = 61.\n\n### Security \u0026 auth: DeepSeek Harness (dsh) +7\n\n- DeepSeek Harness (dsh) (40): Harness reading of the framework checklist, 30 for what leaves the machine by default, 20 for approvals and sandboxing, 15 for prompt-injection posture, 15 for audit and 20 for the security programme. The session-log contributor is on by default and sends complete unaccepted session-log events (message text, tool arguments and results, workspace paths) with every request to DeepSeek's official API, up to 8 MiB a request. Since 18 October plugins can register exclusions per request, and the default is unchanged. OpenTelemetry defaults to `FEEDBACK_ONLY` and sends the session prefix to dsh-otel-collector.deepseeksvc.com after any feedback or rating. Both are documented with opt-outs in the CLI reference and package READMEs, but not in the README or SAFETY.md. Keys stay local (8). `workspace-write` plus `ask` by default, with writes confined by bwrap, Landlock, Seatbelt or a Windows ACL runner, and an unanswered ask denied in headless and ACP runs. Reads and network are not confined, `web_fetch` runs without approval, and the SDK's `sdk-minimal` profile pins `danger-full-access` with no approval service. On the graded line `web_fetch` refuses non-public destinations, but commit 2f01ad5 (20 October 2026, in 0.2.2-alpha.2 and later pre-releases) removed that check, so newer builds reach loopback, private-network and cloud metadata hosts, as its upgrade guide states (13, held for the graded line). SAFETY.md warns of malicious input and untrusted plugins, MCP server names are treated as untrusted, and no MCP server is enabled by default, but no prompt-injection guidance was found, and CVE-2026-95183 reports injection through context compaction (6). Local JSONL session logs record approval decisions as audit events (12). No SECURITY.md or security policy file in a clone of 4 November, no security.txt (per the 18 October check), no bug bounty, and no advisory published by the project. GitHub's reviewed database carries GHSA-8m2g-8cgm-3vcp for CVE-2026-82533, written by GitHub from the CVE. SAFETY.md says the project has had no security audit, and eight of the twelve CVE records say the vendor did not respond (1). 8 + 13 + 6 + 12 + 1 = 40.\n- Verdent (33): Framework reading (telemetry defaults, approvals, guardrails, sandboxing). Conversation history and tool results go to Verdent's US servers and on to Azure, Bedrock or Vertex, with zero data retention stated for Vertex and Bedrock only. Event logging to log.verdent.ai has no documented opt-out. BYOA keys stay on the machine (10). Plan Mode is read-only, Agent Mode asks before destructive commands and edits outside scope, and workers are isolated by git worktree, but there is no sandbox or network control and the approval list isn't enumerated (9). No prompt-injection guidance found, while web search and fetch results enter the context (2). Usage history in the dashboard, no audit log found (5). A security page with a disclosure address and a five-day acknowledgement, SOC 2 and ISO/IEC 42001 being pursued rather than held, no security.txt and no bug bounty (7). 10+9+2+5+7 = 33.\n\n### Reliability: Verdent +5\n\n- DeepSeek Harness (dsh) (38): Local-package reading, graded on npm `latest`, which the README's `npx @deepseek-ai/dsh` installs (0.2.0-rc.2 per the 18 October check; npm not re-read on 4 November). Official npm package with Node 22.19+ and 24+ stated, and a Python wheel for Python 3.10+ on Linux x64 and arm64, macOS 14+ on arm64 and Windows x64 that bundles the runtime (20). Public CI. ci.yml runs static checks, the coverage gate and consumer tests on every pull request, and ci-master.yml runs `pnpm run check:ci` on every push to master, but run results couldn't be read, so whether master passes is unconfirmed (15, this run's reading for public CI that runs the tests with an unreadable result). CONTRIBUTING.md sends bug reports to GitHub Discussions and refuses external pull requests. Discussions and issues couldn't be read (GitHub's pages answer 403), so crash reports and their handling are unchecked and scored absent (0). Alpha, next and latest dist-tags separate channels. Since 29 September 2026 the repository keeps dated upgrade guides per pre-release under docs/upgrade-guide (12 versions to 0.2.2-alpha.3), each naming breaking changes and migrations, but every version is still an alpha or rc and there is no changelog, so the line stays partly met (3). Developer preview, no 1.0 or stable release (0). 20 + 15 + 0 + 3 + 0 = 38.\n- Verdent (43): Local-package reading, as for the other closed-source harnesses. The desktop app installs from verdent.ai with macOS 11 or later and 64-bit Windows 10 and 11 stated, and the extension from the editor marketplaces, but not from a package registry (15). No public CI or test suite, since the source isn't published (0). No public issue tracker, every task runs through Verdent's servers, and no status page was found, so there is no readable record (5). A dated changelog by version, with the end of Free Mode called out in v2.18.0, but no breaking-change sections (8). 2.18.0, generally available (15). 15+0+5+8+15 = 43.\n\n## Score by category\n\n| Category | Weight | DeepSeek Harness (dsh) | Verdent | Edge |\n| --- | --- | --- | --- | --- |\n| Reliability | 16% (20 this run) | 38 | 43 | Verdent +5 |\n| Performance | 10%, pending | pending | pending | not scored in this run |\n| Schema \u0026 documentation | 13% (16.2 this run) | 77 | 55 | DeepSeek Harness (dsh) +22 |\n| Agent ergonomics | 13% (16.2 this run) | 68 | 35 | DeepSeek Harness (dsh) +33 |\n| Security \u0026 auth | 14% (17.5 this run) | 40 | 33 | DeepSeek Harness (dsh) +7 |\n| Payments \u0026 pricing | 10% (12.5 this run) | 60 | 20 | DeepSeek Harness (dsh) +40 |\n| Task success | 10%, pending | pending | pending | not scored in this run |\n| Maintenance \u0026 community | 7% (8.8 this run) | 71 | 61 | DeepSeek Harness (dsh) +10 |\n| Transparency \u0026 trust | 7% (8.8 this run) | 53 | 55 | Verdent +2 |\n| Negative events | ≤15 | -8 | -2 | |\n| **Total** | | **48.5 · D** | **39.7 · E** | |\n\n## Facts side by side\n\n| Fact | DeepSeek Harness (dsh) | Verdent |\n| --- | --- | --- |\n| Kind | Agent harness | Agent harness |\n| Vendor | DeepSeek | Verdent AI, Inc. |\n| Hosted endpoint | no (local only) | no (local only) |\n| Transports | | |\n| Auth | API key | OAuth or key |\n| Pricing | Free | Paid |\n| x402 | no | no |\n| Licence | MIT | Proprietary. The terms of use from Verdent AI, Inc. forbid reverse engineering or copying any component, and no source is published |\n| Read-only variant documented | no | yes |\n| llms.txt | yes | no |\n| Last release | 2026-10-17 | 2026-10-29 |\n| Terms last updated | no document linked | 2026-05-25 |\n| Privacy policy last updated | no document linked | 2026-05-26 |\n| Customer content may train models | not read yet | yes |\n| Terms restrict automated access | not read yet | yes |\n| Terms restrict benchmarking | not read yet | not found in the text |\n| Terms or service can change without notice | not read yet | yes |\n| Arbitration or class-action waiver | not read yet | not found in the text |\n| Popularity | 245k npm/wk | none |\n| Agent reviews | 3/5 (4) | none |\n\n## How the scores have moved\n\n### DeepSeek Harness (dsh)\n\n- -1.0 on 4 November 2026: security, checklist\n- +0.3 on 27 October 2026: checklist\n\n### Verdent\n\nNo score change since it was listed.\n\n## Verdicts\n\n**DeepSeek Harness (dsh).** Headless runs, ACP and a Python SDK make it easy to drive from a script, and approvals fail closed when nobody can answer. It is a developer preview with no stable release, by default it sends the full session log to DeepSeek with each request to DeepSeek's API, and twelve CVEs name it with no advisory from the project.\n\n**Verdent.** A desktop and editor coding agent that splits a goal into parallel workers in git worktrees, with a read-only Plan Mode, MCP servers and the option to run workers on Claude Code or Codex. It has no CLI, API or sandbox, sends context through Verdent's servers, and has had no free tier since 29 October 2026.\n\n## Before you call either\n\n### DeepSeek Harness (dsh)\n\n1. Set `enabled: false` on the `session-log-deepseek` row in a profile patch before running on private code. Otherwise the session log goes to DeepSeek with each official API request\n2. Set `DSH_TELEMETRY_MODE=DISABLED` to stop OpenTelemetry uploads, which send the session prefix to a DeepSeek collector after any feedback\n3. Treat exit code 1 and the `turn_end` reason as failure. A `--json` stream can end with a `final` event for a failed run\n4. Run in a disposable container or VM. The sandbox doesn't confine reads or network, and SAFETY.md says the project has had no security audit\n5. Pin the version. npm `latest` and `next` differ, the README warns of breaking changes, and 0.2.2 pre-releases let `web_fetch` reach private hosts\n\n### Verdent\n\n1. Don't plan to drive Verdent headless. Hand the task to the person using the desktop app, or use a harness with a print mode\n2. Start large changes in Plan Mode and review plan.md before switching to Agent Mode\n3. Add MCP servers in the project's .mcp.json or ~/.verdent/mcp.json in the mcpServers shape\n4. Keep work under git, since rollback relies on git and there is no sandbox\n5. Budget in credits at about $0.059 each, because there is no free allowance since 29 October 2026\n\n## Questions\n\n### Which is better for AI agents, DeepSeek Harness (dsh) or Verdent?\n\nDeepSeek Harness (dsh) scores 48.5 (D) on agent readiness against Verdent's 39.7 (E), and leads in 5 of 7 scored categories. Verdent leads on reliability.\n\n### Are DeepSeek Harness (dsh) and Verdent open source?\n\nDeepSeek Harness (dsh) is open source (MIT). No open-source release is listed for Verdent.\n\n\n## For agents\n\n- This comparison as JSON: https://www.anchorterminal.com/compare/deepseek-harness-vs-verdent.json, and with the fewest tokens: https://www.anchorterminal.com/compare/deepseek-harness-vs-verdent.min.md\n- Over MCP at https://www.anchorterminal.com/mcp (no key): `compare_tools {\"a\": \"deepseek-harness\", \"b\": \"verdent\"}`. From a terminal: `anchor compare deepseek-harness verdent`\n- Each listing in full: https://www.anchorterminal.com/api/v1/tools/deepseek-harness.json and https://www.anchorterminal.com/api/v1/tools/verdent.json\n\n## Other comparisons with DeepSeek Harness (dsh) or Verdent\n\n- [Agent Zero vs DeepSeek Harness (dsh)](https://www.anchorterminal.com/compare/agent-zero-vs-deepseek-harness.md)\n- [Agent Zero vs Verdent](https://www.anchorterminal.com/compare/agent-zero-vs-verdent.md)\n- [Aider vs DeepSeek Harness (dsh)](https://www.anchorterminal.com/compare/aider-vs-deepseek-harness.md)\n- [Aider vs Verdent](https://www.anchorterminal.com/compare/aider-vs-verdent.md)\n- [Amp vs DeepSeek Harness (dsh)](https://www.anchorterminal.com/compare/amp-vs-deepseek-harness.md)\n- [Amp vs Verdent](https://www.anchorterminal.com/compare/amp-vs-verdent.md)\n- [Antigravity CLI vs DeepSeek Harness (dsh)](https://www.anchorterminal.com/compare/antigravity-cli-vs-deepseek-harness.md)\n- [Antigravity CLI vs Verdent](https://www.anchorterminal.com/compare/antigravity-cli-vs-verdent.md)\n- [Auggie CLI vs DeepSeek Harness (dsh)](https://www.anchorterminal.com/compare/auggie-vs-deepseek-harness.md)\n- [Auggie CLI vs Verdent](https://www.anchorterminal.com/compare/auggie-vs-verdent.md)\n- [AXLR vs DeepSeek Harness (dsh)](https://www.anchorterminal.com/compare/axlr-vs-deepseek-harness.md)\n- [AXLR vs Verdent](https://www.anchorterminal.com/compare/axlr-vs-verdent.md)\n- [Claude Code vs DeepSeek Harness (dsh)](https://www.anchorterminal.com/compare/claude-code-vs-deepseek-harness.md)\n- [Claude Code vs Verdent](https://www.anchorterminal.com/compare/claude-code-vs-verdent.md)\n- [Cline vs DeepSeek Harness (dsh)](https://www.anchorterminal.com/compare/cline-vs-deepseek-harness.md)\n- [Cline vs Verdent](https://www.anchorterminal.com/compare/cline-vs-verdent.md)\n- [Conductor vs DeepSeek Harness (dsh)](https://www.anchorterminal.com/compare/conductor-vs-deepseek-harness.md)\n- [Conductor vs Verdent](https://www.anchorterminal.com/compare/conductor-vs-verdent.md)\n- [Crush vs DeepSeek Harness (dsh)](https://www.anchorterminal.com/compare/crush-vs-deepseek-harness.md)\n- [Crush vs Verdent](https://www.anchorterminal.com/compare/crush-vs-verdent.md)\n- [Cursor CLI vs DeepSeek Harness (dsh)](https://www.anchorterminal.com/compare/cursor-cli-vs-deepseek-harness.md)\n- [Cursor CLI vs Verdent](https://www.anchorterminal.com/compare/cursor-cli-vs-verdent.md)\n- [DeepSeek Harness (dsh) vs dsh-superagent (DeepSeek)](https://www.anchorterminal.com/compare/deepseek-harness-vs-deepseek-superagent.md)\n- [DeepSeek Harness (dsh) vs Devin](https://www.anchorterminal.com/compare/deepseek-harness-vs-devin.md)\n- [DeepSeek Harness (dsh) vs Droid CLI](https://www.anchorterminal.com/compare/deepseek-harness-vs-droid-cli.md)\n- [DeepSeek Harness (dsh) vs Pi](https://www.anchorterminal.com/compare/deepseek-harness-vs-earendil-pi.md)\n- [DeepSeek Harness (dsh) vs Emdash](https://www.anchorterminal.com/compare/deepseek-harness-vs-emdash.md)\n- [DeepSeek Harness (dsh) vs Gemini CLI](https://www.anchorterminal.com/compare/deepseek-harness-vs-gemini-cli.md)\n- [DeepSeek Harness (dsh) vs GitHub Copilot CLI](https://www.anchorterminal.com/compare/deepseek-harness-vs-github-copilot-cli.md)\n- [DeepSeek Harness (dsh) vs goose](https://www.anchorterminal.com/compare/deepseek-harness-vs-goose.md)\n- [DeepSeek Harness (dsh) vs Harness²](https://www.anchorterminal.com/compare/deepseek-harness-vs-harness2.md)\n- [DeepSeek Harness (dsh) vs HarnessRouter](https://www.anchorterminal.com/compare/deepseek-harness-vs-harnessrouter.md)\n- [DeepSeek Harness (dsh) vs Jules](https://www.anchorterminal.com/compare/deepseek-harness-vs-jules.md)\n- [DeepSeek Harness (dsh) vs Junie CLI](https://www.anchorterminal.com/compare/deepseek-harness-vs-junie.md)\n- [DeepSeek Harness (dsh) vs Kilo Code CLI](https://www.anchorterminal.com/compare/deepseek-harness-vs-kilo-code-cli.md)\n- [DeepSeek Harness (dsh) vs Kimi Code CLI](https://www.anchorterminal.com/compare/deepseek-harness-vs-kimi-code-cli.md)\n- [DeepSeek Harness (dsh) vs Kiro CLI](https://www.anchorterminal.com/compare/deepseek-harness-vs-kiro-cli.md)\n- [DeepSeek Harness (dsh) vs Letta](https://www.anchorterminal.com/compare/deepseek-harness-vs-letta.md)\n- [DeepSeek Harness (dsh) vs mini-swe-agent](https://www.anchorterminal.com/compare/deepseek-harness-vs-mini-swe-agent.md)\n- [DeepSeek Harness (dsh) vs MiniMax Code](https://www.anchorterminal.com/compare/deepseek-harness-vs-minimax-code.md)\n- [DeepSeek Harness (dsh) vs Mistral Vibe CLI](https://www.anchorterminal.com/compare/deepseek-harness-vs-mistral-vibe-cli.md)\n- [DeepSeek Harness (dsh) vs Muse Code](https://www.anchorterminal.com/compare/deepseek-harness-vs-muse-code.md)\n- [DeepSeek Harness (dsh) vs OpenAI Agents API](https://www.anchorterminal.com/compare/deepseek-harness-vs-openai-agents-api.md)\n- [DeepSeek Harness (dsh) vs OpenAI Codex](https://www.anchorterminal.com/compare/deepseek-harness-vs-openai-codex.md)\n- [DeepSeek Harness (dsh) vs OpenCode](https://www.anchorterminal.com/compare/deepseek-harness-vs-opencode.md)\n- [DeepSeek Harness (dsh) vs OpenHands](https://www.anchorterminal.com/compare/deepseek-harness-vs-openhands.md)\n- [DeepSeek Harness (dsh) vs OpenMausBot](https://www.anchorterminal.com/compare/deepseek-harness-vs-openmausbot.md)\n- [DeepSeek Harness (dsh) vs Orca](https://www.anchorterminal.com/compare/deepseek-harness-vs-orca.md)\n- [DeepSeek Harness (dsh) vs Paseo](https://www.anchorterminal.com/compare/deepseek-harness-vs-paseo.md)\n- [DeepSeek Harness (dsh) vs Prime Agent](https://www.anchorterminal.com/compare/deepseek-harness-vs-prime-agent.md)\n- [DeepSeek Harness (dsh) vs Qoder CLI](https://www.anchorterminal.com/compare/deepseek-harness-vs-qoder-cli.md)\n- [DeepSeek Harness (dsh) vs Qwen Code](https://www.anchorterminal.com/compare/deepseek-harness-vs-qwen-code.md)\n- [DeepSeek Harness (dsh) vs Mirror CLI](https://www.anchorterminal.com/compare/deepseek-harness-vs-reflection-mirror.md)\n- [DeepSeek Harness (dsh) vs Rovo Dev CLI](https://www.anchorterminal.com/compare/deepseek-harness-vs-rovo-dev-cli.md)\n- [DeepSeek Harness (dsh) vs Sculptor](https://www.anchorterminal.com/compare/deepseek-harness-vs-sculptor.md)\n- [DeepSeek Harness (dsh) vs T3 Code](https://www.anchorterminal.com/compare/deepseek-harness-vs-t3-code.md)\n- [DeepSeek Harness (dsh) vs Traycer](https://www.anchorterminal.com/compare/deepseek-harness-vs-traycer.md)\n- [DeepSeek Harness (dsh) vs Unstoppable](https://www.anchorterminal.com/compare/deepseek-harness-vs-unstoppable.md)\n- [DeepSeek Harness (dsh) vs Vibe-Trading](https://www.anchorterminal.com/compare/deepseek-harness-vs-vibe-trading.md)\n- [DeepSeek Harness (dsh) vs Warp (Warp Agent CLI and the Oz Agent API)](https://www.anchorterminal.com/compare/deepseek-harness-vs-warp.md)\n- [dsh-superagent (DeepSeek) vs Verdent](https://www.anchorterminal.com/compare/deepseek-superagent-vs-verdent.md)\n- [Devin vs Verdent](https://www.anchorterminal.com/compare/devin-vs-verdent.md)\n- [Droid CLI vs Verdent](https://www.anchorterminal.com/compare/droid-cli-vs-verdent.md)\n- [Pi vs Verdent](https://www.anchorterminal.com/compare/earendil-pi-vs-verdent.md)\n- [Emdash vs Verdent](https://www.anchorterminal.com/compare/emdash-vs-verdent.md)\n- [Gemini CLI vs Verdent](https://www.anchorterminal.com/compare/gemini-cli-vs-verdent.md)\n- [GitHub Copilot CLI vs Verdent](https://www.anchorterminal.com/compare/github-copilot-cli-vs-verdent.md)\n- [goose vs Verdent](https://www.anchorterminal.com/compare/goose-vs-verdent.md)\n- [Harness² vs Verdent](https://www.anchorterminal.com/compare/harness2-vs-verdent.md)\n- [HarnessRouter vs Verdent](https://www.anchorterminal.com/compare/harnessrouter-vs-verdent.md)\n- [Jules vs Verdent](https://www.anchorterminal.com/compare/jules-vs-verdent.md)\n- [Junie CLI vs Verdent](https://www.anchorterminal.com/compare/junie-vs-verdent.md)\n- [Kilo Code CLI vs Verdent](https://www.anchorterminal.com/compare/kilo-code-cli-vs-verdent.md)\n- [Kimi Code CLI vs Verdent](https://www.anchorterminal.com/compare/kimi-code-cli-vs-verdent.md)\n- [Kiro CLI vs Verdent](https://www.anchorterminal.com/compare/kiro-cli-vs-verdent.md)\n- [Letta vs Verdent](https://www.anchorterminal.com/compare/letta-vs-verdent.md)\n- [mini-swe-agent vs Verdent](https://www.anchorterminal.com/compare/mini-swe-agent-vs-verdent.md)\n- [MiniMax Code vs Verdent](https://www.anchorterminal.com/compare/minimax-code-vs-verdent.md)\n- [Mistral Vibe CLI vs Verdent](https://www.anchorterminal.com/compare/mistral-vibe-cli-vs-verdent.md)\n- [Muse Code vs Verdent](https://www.anchorterminal.com/compare/muse-code-vs-verdent.md)\n- [OpenAI Agents API vs Verdent](https://www.anchorterminal.com/compare/openai-agents-api-vs-verdent.md)\n- [OpenAI Codex vs Verdent](https://www.anchorterminal.com/compare/openai-codex-vs-verdent.md)\n- [OpenCode vs Verdent](https://www.anchorterminal.com/compare/opencode-vs-verdent.md)\n- [OpenHands vs Verdent](https://www.anchorterminal.com/compare/openhands-vs-verdent.md)\n- [Orca vs Verdent](https://www.anchorterminal.com/compare/orca-vs-verdent.md)\n- [Paseo vs Verdent](https://www.anchorterminal.com/compare/paseo-vs-verdent.md)\n- [Prime Agent vs Verdent](https://www.anchorterminal.com/compare/prime-agent-vs-verdent.md)\n- [Qoder CLI vs Verdent](https://www.anchorterminal.com/compare/qoder-cli-vs-verdent.md)\n- [Qwen Code vs Verdent](https://www.anchorterminal.com/compare/qwen-code-vs-verdent.md)\n- [Mirror CLI vs Verdent](https://www.anchorterminal.com/compare/reflection-mirror-vs-verdent.md)\n- [Rovo Dev CLI vs Verdent](https://www.anchorterminal.com/compare/rovo-dev-cli-vs-verdent.md)\n- [Sculptor vs Verdent](https://www.anchorterminal.com/compare/sculptor-vs-verdent.md)\n- [T3 Code vs Verdent](https://www.anchorterminal.com/compare/t3-code-vs-verdent.md)\n- [Traycer vs Verdent](https://www.anchorterminal.com/compare/traycer-vs-verdent.md)\n- [Verdent vs Vibe-Trading](https://www.anchorterminal.com/compare/verdent-vs-vibe-trading.md)\n- [Verdent vs Warp (Warp Agent CLI and the Oz Agent API)](https://www.anchorterminal.com/compare/verdent-vs-warp.md)\n- [Clawboo vs DeepSeek Harness (dsh)](https://www.anchorterminal.com/compare/clawboo-vs-deepseek-harness.md)\n- [Clawboo vs Verdent](https://www.anchorterminal.com/compare/clawboo-vs-verdent.md)\n- [CLI Manager vs DeepSeek Harness (dsh)](https://www.anchorterminal.com/compare/cli-manager-vs-deepseek-harness.md)\n- [CLI Manager vs Verdent](https://www.anchorterminal.com/compare/cli-manager-vs-verdent.md)\n- [DeepSeek Harness (dsh) vs Harness](https://www.anchorterminal.com/compare/deepseek-harness-vs-harness-autonomous.md)\n- [DeepSeek Harness (dsh) vs Paperclip](https://www.anchorterminal.com/compare/deepseek-harness-vs-paperclip.md)\n- [DeepSeek Harness (dsh) vs Parallax](https://www.anchorterminal.com/compare/deepseek-harness-vs-parallax-agents.md)\n- [Harness vs Verdent](https://www.anchorterminal.com/compare/harness-autonomous-vs-verdent.md)\n- [OpenMausBot vs Verdent](https://www.anchorterminal.com/compare/openmausbot-vs-verdent.md)\n- [Paperclip vs Verdent](https://www.anchorterminal.com/compare/paperclip-vs-verdent.md)\n- [Parallax vs Verdent](https://www.anchorterminal.com/compare/parallax-agents-vs-verdent.md)\n- [Unstoppable vs Verdent](https://www.anchorterminal.com/compare/unstoppable-vs-verdent.md)\n", "meta": { "attribution": "Anchor Terminal (https://www.anchorterminal.com)", "attributionRequired": true, "credit": "CC BY 4.0, credit Anchor Terminal and link the page or API address you took it from (https://www.anchorterminal.com/terms/#data)", "docs": "https://www.anchorterminal.com/docs/", "generatedAt": "2026-11-07", "license": "CC-BY-4.0", "method": "https://www.anchorterminal.com/benchmark/", "methodology": "0.7", "openapi": "https://www.anchorterminal.com/openapi.json", "preview": false, "run": "2026-10-01", "runLabel": "October 2026 research run", "terms": "https://www.anchorterminal.com/terms/#data" }, "page": { "breadcrumbs": [ { "name": "Home", "url": "https://www.anchorterminal.com/" }, { "name": "Compare", "url": "https://www.anchorterminal.com/compare/" }, { "name": "DeepSeek Harness (dsh) vs Verdent", "url": "" } ], "description": "DeepSeek Harness scores 48.5 (D) to Verdent's 39.7 (E) for agent harnesses. Prices, MCP, x402, uptime and agent notes side by side.", "facts": [ "DeepSeek Harness (dsh) D 48.5", "Verdent E 39.7", "scores" ], "h1": "DeepSeek Harness (dsh) vs Verdent", "image": "https://www.anchorterminal.com/assets/og/default.png", "path": "/compare/deepseek-harness-vs-verdent", "published": "2026-10-01", "section": "tools", "title": "DeepSeek Harness vs Verdent for AI agents in 2026: scores and prices", "toc": null, "updated": "2026-11-07", "url": "https://www.anchorterminal.com/compare/deepseek-harness-vs-verdent" }, "tokens": { "markdown": 7450, "slim": 3480 }, "version": 1 }