{
  "fixes": {
    "slug": "nanoclaw",
    "name": "NanoClaw",
    "listing": "https://www.anchorterminal.com/tools/nanoclaw",
    "markdown": "# Fix list: NanoClaw\n\nFrom Anchor Terminal's listing at https://www.anchorterminal.com/tools/nanoclaw, the October 2026 research run, assessed 12 October 2026. Grade B, 68.3 out of 100.\n\nThis is everything the published grade says the listing lacks, the biggest possible gain to the total first. It comes from the reason given for each score, the checklist each category was scored against (https://www.anchorterminal.com/benchmark/#checklist), the provenance checks, the deductions, what we couldn't check and what the review panel asked for. A fix counts at the next check, once it's public.\n\nFor a coding agent working on NanoClaw: work through the items below in the product, its docs and its public pages. Each category gives the reason for its score, with the points each checklist item earned, and the checklist itself, so the gap is the items that earned less than their points. Change the product, not the wording, and keep a note of what you changed and where it's published.\n\n## 1. Payments \u0026 pricing, 60 out of 100, up to 5 more on the total\n\nWhy it scored 60: No payment protocol (0). Free under the MIT licence with nothing to buy, so the self-hosted rule gives 20, 20 and 20 on the last three lines. The optional account is free. The owner still needs a model credential, or a local model through the Ollama provider skill.\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-payments):\n\nThe published rubric, also on the [x402 page](https://www.anchorterminal.com/x402/).\n\n- 40, a machine payment protocol (x402, MPP or L402) on the tool's own endpoints. 10 to 30 when it covers only some endpoints or only goes through a third party, and the note says which.\n- 20, per-call or per-unit pricing published without a login. 10 for public plan-only pricing, 0 for \"contact sales\" or prices behind a login.\n- 20, a free tier or trial that doesn't need a card.\n- 20, autonomous onboarding, meaning an agent can get access without a person signing up in a browser (keyless use, x402, a programmatic key API).\n\nPayment platforms and agent wallets rarely charge for their own API over a machine protocol, so the first line has steps for them, and the highest one that applies counts. 40 when x402, MPP or L402 runs on all their own endpoints, 30 when it runs on part of their own API, 25 when their merchants can accept one, 20 for running a facilitator, 15 for paying as a buyer, and 0 when the only protocol is their own. Merchant acceptance sits above a facilitator because the platform's own customers can charge agents through it, while a facilitator settles for sellers who wire up the protocol themselves. The counter-argument (a facilitator does more for the protocol as a whole) has a point. Each note says which step applied.\n\nOpen-source software you run yourself is scored on its hosted or paid option if it has one. A free, self-hosted package with nothing to buy gets 20, 20 and 20 for the last three lines, and 0 to 40 for the first only if it ships a payment protocol.\n\n## 2. Agent ergonomics, 70 out of 100, up to 4.9 more on the total\n\nWhy it scored 70: Harness reading, scored on what an owner or pipeline driving it has to supply. The built-in MCP server has nine tools beside 16 allowed Claude Code tools, and the two memory files loaded into each new context are capped at 16,000 characters each (21). A 2,048 process limit is on by default, CPU and memory limits are unset, and no spending or turn cap was found (11). The `ncl` CLI and a local CLI channel work over a Unix socket, with no HTTP interface (12). Sessions, approvals and delivery retries persist in SQLite across host restarts, and the updater takes a snapshot and rolls back on failure (18). One script installs it, but it needs Docker, Node and a Claude or Codex credential, channel skills need Claude Code, and changing behaviour means editing a fork. No SDK (8).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-ergonomics):\n\n- 0 to 25, context cost. For MCP, the number and size of the tool definitions (25 for ten or fewer compact tools, 15 for 11 to 30, 5 for more than 30, plus up to 10 back for toolsets, dynamic loading or read-only subsets). For APIs, whether responses can be sized (field selection, limits, summaries).\n- 20, pagination, filtering and output-size controls.\n- 20, actionable, documented error responses, codes and messages an agent can recover from.\n- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.\n- 15, sensible defaults, few required parameters, and official SDKs in at least two languages.\n\nModels are read for tool use, structured output, prompt caching, context length, batch and SDKs. Frameworks for how much code and how many defaults a tool-calling agent with MCP needs.\n\n## 3. Security \u0026 auth, 74 out of 100, up to 4.6 more on the total\n\nWhy it scored 74: 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. Setup events go to PostHog under a random install id unless `NANOCLAW_NO_DIAGNOSTICS=1` is set, which the README discloses, and no runtime telemetry was found in the host or container source. Message content goes to the model provider chosen (22). Sessions run in non-root containers with all capabilities dropped, extra mounts are denied without an allowlist, and API keys stay in a gateway on the host. Self-modification, new chats and `ncl` writes wait for an admin. But the Claude provider runs with permission prompts off, sending to wired chats needs no approval, per-request credential approval is opt-in and egress lockdown is off by default (15). The security page assumes injection succeeds and lists what a manipulated agent can still do. Sender policies and a command gate limit who can reach an agent, and no content screening was found (13). A dropped-messages log, an approvals list, task run logs and session history through `ncl`, and gateway activity in the OneCLI dashboard (11). SECURITY.md with private reporting, fixes credited to reporters in the changelog, signed image verification and pinned Actions in CI. The eight advisories the 9 October changelog links were not public on 12 October, and there is no security.txt and no bounty (13).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-security):\n\n- 0 to 30, the credential model. 30 for OAuth 2.1 with scopes, or scoped and revocable keys with rotation. 20 for plain revocable API keys. 10 for one all-powerful key. 10 off when a secret can travel in a URL query string as a documented option.\n- 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions.\n- 0 to 15, prompt-injection posture where the tool returns untrusted content (documented mitigations or guidance). A tool that returns no untrusted content gets 10.\n- 0 to 15, audit logs or per-call visibility for the operator.\n- 0 to 20, a security programme. security.txt or a disclosure policy, a bug bounty, SOC 2 or ISO 27001, advisories handled in public.\n\nModels are read for retention, whether API data trains models (and whether that's off by default), zero-retention options and certifications. Frameworks for telemetry defaults, approval hooks, guardrails and sandboxing.\n\n## 4. Reliability, 78 out of 100, up to 4.4 more on the total\n\nWhy it scored 78: Read as software the owner runs. Installed by `git clone` and `bash nanoclaw.sh`, with no registry package. macOS, Linux and WSL2 with Docker are stated, but the home page says Node.js 20 or later and the README and `package.json` say 22 or later (13). The CI workflow runs formatting, type checks and host and container tests on Node 22 and 24, and the 14 completed runs on main from 5 to 11 October 2026 all passed (25). 294 open issues and 750 open pull requests on 12 October 2026, and 10 of the 12 newest open issues had no reply, among them a report that updates do not rebuild per-group images (12). The changelog marks breaking changes and opens each release with steps to take before updating, and versions moved from 2.4.0 to calendar numbering on 9 October 2026 (13). Version 2 since 22 April 2026, with stable, beta and edge update channels (15).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability):\n\nHosted APIs, MCP servers, models and platforms.\n\n- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).\n- 0 to 30, the incident record for the last 90 days on that page. 30 for a clean record or trivial incidents only, 20 for minor incidents only, 10 for one major outage (an hour or more of a core API down, or errors across the board), 0 for several. 5 when there's no history we could read, and the note says so.\n- 15, rate limits documented with numbers.\n- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.\n- 10, an SLA published for any paid tier.\n- 10, the surface agents use is generally available, not beta or preview.\n\nLocal packages, SDKs, frameworks and stdio MCP servers.\n\n- 20, installs from an official package with supported runtimes stated.\n- 25, a public CI and test suite, passing on the default branch.\n- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).\n- 15, semver discipline and breaking changes called out in a changelog.\n- 15, version 1.0 or later, or declared stable.\n\nProtocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors.\n\n## 5. Transparency \u0026 trust, 65 out of 100, up to 3.1 more on the total\n\nMade of editorial 77, provenance 53.\n\nWhy it scored 65: MIT (30). The README has a section on what leaves the machine, and the privacy policy describes the optional account in detail, with IP addresses cut to a /24 network. The policy says it does not cover the software and says NanoClaw does not report back, without mentioning the setup events sent to PostHog, and gives no retention period for them (19). Breaking changes are marked with migration steps and a written release process, with no stated notice period (13). Setup diagnostics are disclosed in the README with an opt-out variable, and are on by default (15).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-transparency):\n\n- 0 to 30, source availability and licence clarity. 30 for open source under an OSI licence, 15 for closed with clear terms, 0 for unclear terms.\n- 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors).\n- 0 to 20, a deprecation policy or notices with dates.\n- 0 to 20, telemetry disclosed with an opt-out (local software), or subprocessors and data locations disclosed (hosted).\n\nThe other half of Transparency and trust is the provenance score, computed from checked facts (below). The category score is the mean of the two.\n\nProvenance checks not met in full (half of this category, computed from checked facts):\n\n- Domain age: nanoclaw.dev, no registry record we could read (0 of 15)\n- Status page: not found (0 of 10)\n- security.txt: not found (0 of 10)\n\n## 6. Schema \u0026 documentation, 82 out of 100, up to 2.9 more on the total\n\nWhy it scored 82: Harness reading, as for the agent programs we list. No HTTP API and no OpenAPI file. The docs carry references for the nine built-in MCP tools, the `ncl` CLI, the database schema, container configuration and the template format, and the tools declare JSON Schema inputs in the source (17). llms.txt at docs.nanoclaw.dev with a Markdown twin of every page (10). The security page says what each layer stops and lists four things it does not, and the hardening page gives the default of each setting (17). Tool parameters are tabled with types and required flags, and sender policies are a fixed set of four values, though message content is free text (11). Step-by-step guides and a troubleshooting page, with no error reference found (12). A dated CHANGELOG.md, mirrored on the docs site, and a written release process (15).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-schema):\n\nAPIs and MCP servers.\n\n- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).\n- 10, llms.txt or Markdown docs served for agents.\n- 0 to 20, descriptions that say what a tool is for, when to use it and when not to, read from the tool definitions in the source or the API reference.\n- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.\n- 0 to 15, examples and documented error responses.\n- 15, versioning and a public changelog.\n\nModels are read from the API reference, the OpenAPI file, llms.txt, the structured-output and tool-use docs and the model cards. Frameworks from docs a model can follow, typed interfaces, examples and the API reference.\n\n## 7. Maintenance \u0026 community, 78 out of 100, up to 1.9 more on the total\n\nWhy it scored 78: 2026.10.0 on 9 October 2026 (30). Five stable releases in the 90 days to 12 October, from 2.1.54 to 2026.10.0, and two release candidates (20). Reported security flaws were fixed and credited, but 294 issues and 750 pull requests were open, and 10 of the 12 newest open issues had no reply (11). No registry package. Channels and providers ship from registry branches through skills, and a prebuilt agent image is pinned by digest and checked with cosign in CI (8). CI on Node 22 and 24 with frozen lockfiles, Actions pinned to commit hashes and a three-day release-age gate on dependencies, which the 2.4.0 notes say had been ignored until 23 September 2026 (9).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-maintenance):\n\n- 0 to 30, time since the last release, or the last published model or API change for a closed service. 30 within 30 days, 20 within 90, 10 within 180, 0 older.\n- 20, at least three releases or dated changelog entries in the last 90 days.\n- 0 to 25, responsiveness. Issues and pull requests answered on GitHub (the open issues and how recent the replies are). For closed services, a public changelog and a support or community channel that answers, 0 to 15.\n- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).\n- 10, package health, current dependencies and CI.\n\nModels are read for deprecation notice periods and model churn rather than release counts.\n\n## Deductions\n\nEach comes off the total. A fixed and documented problem counts for less at the next check.\n\n- 2026-10-09. Release 2026.10.0 fixed four symlink flaws through which a manipulated agent could make the host read, write or delete files outside its folders, plus an admin-command bypass, a cross-group read through `ncl` and a command injection in Mattermost setup. Fixed and described in the changelog within the week, but the advisories it links (GHSA-frvf-5rf5-89fq and seven others) were not public on 12 October 2026. -5. https://github.com/nanocoai/nanoclaw/blob/main/CHANGELOG.md\n\n## What we couldn't check\n\nWhat we couldn't read counted as absent. Publishing it on a page a plain HTTP fetch can read (not only in a browser) lets the next check count it.\n\n- Kind is recorded as harness, the closest kind the briefs define for an agent program the owner runs. NanoClaw has no HTTP API and is not an MCP server\n- `assistant.accounts` is claimed on the credential gateway, which holds API keys and OAuth connections per agent group. Trunk has no first-party mail or calendar integration, so a stricter reading would drop it\n- Why the eight advisories linked from the 2026.10.0 changelog were not public three days after the release. They may be unpublished drafts\n- Whether setup shows a consent prompt for diagnostics. We found only the `NANOCLAW_NO_DIAGNOSTICS` check in the setup scripts, and the README calls the diagnostics anonymous\n- What the OneCLI gateway flaw fixed in 1.42.0 allowed. The changelog calls it a credential-injection bypass and we did not read OneCLI's own notes\n- unchecked: how the OneCLI Agent Vault stores secrets at rest. We read NanoClaw's docs and not OneCLI's\n- unchecked: portal.nanoclaw.dev, nanoco.ai, the blog and the registration date of nanoclaw.dev\n- unchecked: what a month of use costs. No figure is published and no model provider's price page was read\n- unchecked: reply times on issues beyond the 50 newest open items, and the state of the channels and providers branches\n- The first-release date was not established. The repository was created on 31 January 2026 and the changelog's version 2.0.0 is dated 22 April 2026\n- github.com's robots.txt closes tree, tags and commit pages to Anthropic agents, so the source was read from a shallow clone and blob pages, and release dates from the API. api.github.com answered 404 for robots.txt, read as no rules\n\n## Weaknesses\n\n- The Claude provider runs with `permissionMode: 'bypassPermissions'`, so shell, file and web tools run inside the container with no prompt\n- Egress lockdown is off by default, so a container can reach the internet directly unless `NANOCLAW_EGRESS_LOCKDOWN=true` is set\n- The 2026.10.0 changelog links eight advisories on the repository, and on 12 October 2026 the advisories page listed none and one linked advisory returned 404\n- Setup sends events to PostHog unless `NANOCLAW_NO_DIAGNOSTICS=1` is set, and the privacy policy does not mention them\n- The docs say trunk has no Gmail or Google Calendar integration, while the GitHub repository description still names Gmail\n- 294 open issues and 750 open pull requests on 12 October 2026, and 10 of the 12 newest open issues had no reply\n\n## What costs an agent a turn today\n\nThe notes we give agents before they call it. Each one is a workaround an agent shouldn't need.\n\n- Install with `git clone` and `bash nanoclaw.sh`. There is no registry package, and setup needs Docker, Node 22 or later and a Claude or Codex credential\n- Set `NANOCLAW_EGRESS_LOCKDOWN=true` before giving an agent credentials or an audience of strangers. The default leaves egress open\n- Put chats with different audiences in separate agent groups. Every session of a group shares one read-write workspace and memory tree\n- Turn on per-request approval for a secret in the gateway when its use should wait for a person. It is off unless configured\n- On OneCLI installs, move the gateway to 1.42.0 by the upgrade guide before updating to 2026.10.0. The updater stops until that is done\n\n## When it's done\n\nSend what changed and where it's published as a dispute (https://www.anchorterminal.com/builders/#disputes, or `POST https://www.anchorterminal.com/api/v1/contact` with `\"kind\": \"dispute\"`). Disputes are answered in public, and the listing is checked again by the same checklist. Paying for an audit or a listing claim changes nothing here.\n",
    "grade": "B",
    "score": 68.3,
    "assessed": "2026-10-12",
    "run": "October 2026 research run",
    "categories": [
      {
        "key": "payments",
        "name": "Payments \u0026 pricing",
        "score": 60,
        "maxGain": 5,
        "reason": "No payment protocol (0). Free under the MIT licence with nothing to buy, so the self-hosted rule gives 20, 20 and 20 on the last three lines. The optional account is free. The owner still needs a model credential, or a local model through the Ollama provider skill.",
        "checklist": [
          "The published rubric, also on the [x402 page](https://www.anchorterminal.com/x402/).",
          "- 40, a machine payment protocol (x402, MPP or L402) on the tool's own endpoints. 10 to 30 when it covers only some endpoints or only goes through a third party, and the note says which.\n- 20, per-call or per-unit pricing published without a login. 10 for public plan-only pricing, 0 for \"contact sales\" or prices behind a login.\n- 20, a free tier or trial that doesn't need a card.\n- 20, autonomous onboarding, meaning an agent can get access without a person signing up in a browser (keyless use, x402, a programmatic key API).",
          "Payment platforms and agent wallets rarely charge for their own API over a machine protocol, so the first line has steps for them, and the highest one that applies counts. 40 when x402, MPP or L402 runs on all their own endpoints, 30 when it runs on part of their own API, 25 when their merchants can accept one, 20 for running a facilitator, 15 for paying as a buyer, and 0 when the only protocol is their own. Merchant acceptance sits above a facilitator because the platform's own customers can charge agents through it, while a facilitator settles for sellers who wire up the protocol themselves. The counter-argument (a facilitator does more for the protocol as a whole) has a point. Each note says which step applied.",
          "Open-source software you run yourself is scored on its hosted or paid option if it has one. A free, self-hosted package with nothing to buy gets 20, 20 and 20 for the last three lines, and 0 to 40 for the first only if it ships a payment protocol."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-payments"
      },
      {
        "key": "ergonomics",
        "name": "Agent ergonomics",
        "score": 70,
        "maxGain": 4.9,
        "reason": "Harness reading, scored on what an owner or pipeline driving it has to supply. The built-in MCP server has nine tools beside 16 allowed Claude Code tools, and the two memory files loaded into each new context are capped at 16,000 characters each (21). A 2,048 process limit is on by default, CPU and memory limits are unset, and no spending or turn cap was found (11). The `ncl` CLI and a local CLI channel work over a Unix socket, with no HTTP interface (12). Sessions, approvals and delivery retries persist in SQLite across host restarts, and the updater takes a snapshot and rolls back on failure (18). One script installs it, but it needs Docker, Node and a Claude or Codex credential, channel skills need Claude Code, and changing behaviour means editing a fork. No SDK (8).",
        "checklist": [
          "- 0 to 25, context cost. For MCP, the number and size of the tool definitions (25 for ten or fewer compact tools, 15 for 11 to 30, 5 for more than 30, plus up to 10 back for toolsets, dynamic loading or read-only subsets). For APIs, whether responses can be sized (field selection, limits, summaries).\n- 20, pagination, filtering and output-size controls.\n- 20, actionable, documented error responses, codes and messages an agent can recover from.\n- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.\n- 15, sensible defaults, few required parameters, and official SDKs in at least two languages.",
          "Models are read for tool use, structured output, prompt caching, context length, batch and SDKs. Frameworks for how much code and how many defaults a tool-calling agent with MCP needs."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-ergonomics"
      },
      {
        "key": "security",
        "name": "Security \u0026 auth",
        "score": 74,
        "maxGain": 4.6,
        "reason": "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. Setup events go to PostHog under a random install id unless `NANOCLAW_NO_DIAGNOSTICS=1` is set, which the README discloses, and no runtime telemetry was found in the host or container source. Message content goes to the model provider chosen (22). Sessions run in non-root containers with all capabilities dropped, extra mounts are denied without an allowlist, and API keys stay in a gateway on the host. Self-modification, new chats and `ncl` writes wait for an admin. But the Claude provider runs with permission prompts off, sending to wired chats needs no approval, per-request credential approval is opt-in and egress lockdown is off by default (15). The security page assumes injection succeeds and lists what a manipulated agent can still do. Sender policies and a command gate limit who can reach an agent, and no content screening was found (13). A dropped-messages log, an approvals list, task run logs and session history through `ncl`, and gateway activity in the OneCLI dashboard (11). SECURITY.md with private reporting, fixes credited to reporters in the changelog, signed image verification and pinned Actions in CI. The eight advisories the 9 October changelog links were not public on 12 October, and there is no security.txt and no bounty (13).",
        "checklist": [
          "- 0 to 30, the credential model. 30 for OAuth 2.1 with scopes, or scoped and revocable keys with rotation. 20 for plain revocable API keys. 10 for one all-powerful key. 10 off when a secret can travel in a URL query string as a documented option.\n- 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions.\n- 0 to 15, prompt-injection posture where the tool returns untrusted content (documented mitigations or guidance). A tool that returns no untrusted content gets 10.\n- 0 to 15, audit logs or per-call visibility for the operator.\n- 0 to 20, a security programme. security.txt or a disclosure policy, a bug bounty, SOC 2 or ISO 27001, advisories handled in public.",
          "Models are read for retention, whether API data trains models (and whether that's off by default), zero-retention options and certifications. Frameworks for telemetry defaults, approval hooks, guardrails and sandboxing."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-security"
      },
      {
        "key": "reliability",
        "name": "Reliability",
        "score": 78,
        "maxGain": 4.4,
        "reason": "Read as software the owner runs. Installed by `git clone` and `bash nanoclaw.sh`, with no registry package. macOS, Linux and WSL2 with Docker are stated, but the home page says Node.js 20 or later and the README and `package.json` say 22 or later (13). The CI workflow runs formatting, type checks and host and container tests on Node 22 and 24, and the 14 completed runs on main from 5 to 11 October 2026 all passed (25). 294 open issues and 750 open pull requests on 12 October 2026, and 10 of the 12 newest open issues had no reply, among them a report that updates do not rebuild per-group images (12). The changelog marks breaking changes and opens each release with steps to take before updating, and versions moved from 2.4.0 to calendar numbering on 9 October 2026 (13). Version 2 since 22 April 2026, with stable, beta and edge update channels (15).",
        "checklist": [
          "Hosted APIs, MCP servers, models and platforms.",
          "- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).\n- 0 to 30, the incident record for the last 90 days on that page. 30 for a clean record or trivial incidents only, 20 for minor incidents only, 10 for one major outage (an hour or more of a core API down, or errors across the board), 0 for several. 5 when there's no history we could read, and the note says so.\n- 15, rate limits documented with numbers.\n- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.\n- 10, an SLA published for any paid tier.\n- 10, the surface agents use is generally available, not beta or preview.",
          "Local packages, SDKs, frameworks and stdio MCP servers.",
          "- 20, installs from an official package with supported runtimes stated.\n- 25, a public CI and test suite, passing on the default branch.\n- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).\n- 15, semver discipline and breaking changes called out in a changelog.\n- 15, version 1.0 or later, or declared stable.",
          "Protocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-reliability"
      },
      {
        "key": "transparency",
        "name": "Transparency \u0026 trust",
        "score": 65,
        "maxGain": 3.1,
        "reason": "MIT (30). The README has a section on what leaves the machine, and the privacy policy describes the optional account in detail, with IP addresses cut to a /24 network. The policy says it does not cover the software and says NanoClaw does not report back, without mentioning the setup events sent to PostHog, and gives no retention period for them (19). Breaking changes are marked with migration steps and a written release process, with no stated notice period (13). Setup diagnostics are disclosed in the README with an opt-out variable, and are on by default (15).",
        "blend": "editorial 77, provenance 53",
        "checklist": [
          "- 0 to 30, source availability and licence clarity. 30 for open source under an OSI licence, 15 for closed with clear terms, 0 for unclear terms.\n- 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors).\n- 0 to 20, a deprecation policy or notices with dates.\n- 0 to 20, telemetry disclosed with an opt-out (local software), or subprocessors and data locations disclosed (hosted).",
          "The other half of Transparency and trust is the provenance score, computed from checked facts (below). The category score is the mean of the two."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-transparency"
      },
      {
        "key": "schema",
        "name": "Schema \u0026 documentation",
        "score": 82,
        "maxGain": 2.9,
        "reason": "Harness reading, as for the agent programs we list. No HTTP API and no OpenAPI file. The docs carry references for the nine built-in MCP tools, the `ncl` CLI, the database schema, container configuration and the template format, and the tools declare JSON Schema inputs in the source (17). llms.txt at docs.nanoclaw.dev with a Markdown twin of every page (10). The security page says what each layer stops and lists four things it does not, and the hardening page gives the default of each setting (17). Tool parameters are tabled with types and required flags, and sender policies are a fixed set of four values, though message content is free text (11). Step-by-step guides and a troubleshooting page, with no error reference found (12). A dated CHANGELOG.md, mirrored on the docs site, and a written release process (15).",
        "checklist": [
          "APIs and MCP servers.",
          "- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).\n- 10, llms.txt or Markdown docs served for agents.\n- 0 to 20, descriptions that say what a tool is for, when to use it and when not to, read from the tool definitions in the source or the API reference.\n- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.\n- 0 to 15, examples and documented error responses.\n- 15, versioning and a public changelog.",
          "Models are read from the API reference, the OpenAPI file, llms.txt, the structured-output and tool-use docs and the model cards. Frameworks from docs a model can follow, typed interfaces, examples and the API reference."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-schema"
      },
      {
        "key": "maintenance",
        "name": "Maintenance \u0026 community",
        "score": 78,
        "maxGain": 1.9,
        "reason": "2026.10.0 on 9 October 2026 (30). Five stable releases in the 90 days to 12 October, from 2.1.54 to 2026.10.0, and two release candidates (20). Reported security flaws were fixed and credited, but 294 issues and 750 pull requests were open, and 10 of the 12 newest open issues had no reply (11). No registry package. Channels and providers ship from registry branches through skills, and a prebuilt agent image is pinned by digest and checked with cosign in CI (8). CI on Node 22 and 24 with frozen lockfiles, Actions pinned to commit hashes and a three-day release-age gate on dependencies, which the 2.4.0 notes say had been ignored until 23 September 2026 (9).",
        "checklist": [
          "- 0 to 30, time since the last release, or the last published model or API change for a closed service. 30 within 30 days, 20 within 90, 10 within 180, 0 older.\n- 20, at least three releases or dated changelog entries in the last 90 days.\n- 0 to 25, responsiveness. Issues and pull requests answered on GitHub (the open issues and how recent the replies are). For closed services, a public changelog and a support or community channel that answers, 0 to 15.\n- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).\n- 10, package health, current dependencies and CI.",
          "Models are read for deprecation notice periods and model churn rather than release counts."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-maintenance"
      }
    ],
    "provenance": [
      {
        "label": "Domain age",
        "value": "nanoclaw.dev, no registry record we could read",
        "points": 0,
        "max": 15
      },
      {
        "label": "Status page",
        "value": "not found",
        "points": 0,
        "max": 10
      },
      {
        "label": "security.txt",
        "value": "not found",
        "points": 0,
        "max": 10
      }
    ],
    "deductions": [
      "2026-10-09. Release 2026.10.0 fixed four symlink flaws through which a manipulated agent could make the host read, write or delete files outside its folders, plus an admin-command bypass, a cross-group read through `ncl` and a command injection in Mattermost setup. Fixed and described in the changelog within the week, but the advisories it links (GHSA-frvf-5rf5-89fq and seven others) were not public on 12 October 2026. -5. https://github.com/nanocoai/nanoclaw/blob/main/CHANGELOG.md"
    ],
    "unchecked": [
      "Kind is recorded as harness, the closest kind the briefs define for an agent program the owner runs. NanoClaw has no HTTP API and is not an MCP server",
      "`assistant.accounts` is claimed on the credential gateway, which holds API keys and OAuth connections per agent group. Trunk has no first-party mail or calendar integration, so a stricter reading would drop it",
      "Why the eight advisories linked from the 2026.10.0 changelog were not public three days after the release. They may be unpublished drafts",
      "Whether setup shows a consent prompt for diagnostics. We found only the `NANOCLAW_NO_DIAGNOSTICS` check in the setup scripts, and the README calls the diagnostics anonymous",
      "What the OneCLI gateway flaw fixed in 1.42.0 allowed. The changelog calls it a credential-injection bypass and we did not read OneCLI's own notes",
      "unchecked: how the OneCLI Agent Vault stores secrets at rest. We read NanoClaw's docs and not OneCLI's",
      "unchecked: portal.nanoclaw.dev, nanoco.ai, the blog and the registration date of nanoclaw.dev",
      "unchecked: what a month of use costs. No figure is published and no model provider's price page was read",
      "unchecked: reply times on issues beyond the 50 newest open items, and the state of the channels and providers branches",
      "The first-release date was not established. The repository was created on 31 January 2026 and the changelog's version 2.0.0 is dated 22 April 2026",
      "github.com's robots.txt closes tree, tags and commit pages to Anthropic agents, so the source was read from a shallow clone and blob pages, and release dates from the API. api.github.com answered 404 for robots.txt, read as no rules"
    ],
    "weaknesses": [
      "The Claude provider runs with `permissionMode: 'bypassPermissions'`, so shell, file and web tools run inside the container with no prompt",
      "Egress lockdown is off by default, so a container can reach the internet directly unless `NANOCLAW_EGRESS_LOCKDOWN=true` is set",
      "The 2026.10.0 changelog links eight advisories on the repository, and on 12 October 2026 the advisories page listed none and one linked advisory returned 404",
      "Setup sends events to PostHog unless `NANOCLAW_NO_DIAGNOSTICS=1` is set, and the privacy policy does not mention them",
      "The docs say trunk has no Gmail or Google Calendar integration, while the GitHub repository description still names Gmail",
      "294 open issues and 750 open pull requests on 12 October 2026, and 10 of the 12 newest open issues had no reply"
    ],
    "agentNotes": [
      "Install with `git clone` and `bash nanoclaw.sh`. There is no registry package, and setup needs Docker, Node 22 or later and a Claude or Codex credential",
      "Set `NANOCLAW_EGRESS_LOCKDOWN=true` before giving an agent credentials or an audience of strangers. The default leaves egress open",
      "Put chats with different audiences in separate agent groups. Every session of a group shares one read-write workspace and memory tree",
      "Turn on per-request approval for a secret in the gateway when its use should wait for a person. It is off unless configured",
      "On OneCLI installs, move the gateway to 1.42.0 by the upgrade guide before updating to 2026.10.0. The updater stops until that is done"
    ],
    "recheck": "https://www.anchorterminal.com/builders/#disputes"
  },
  "meta": {
    "attribution": "Anchor Terminal (https://www.anchorterminal.com)",
    "docs": "https://www.anchorterminal.com/docs/",
    "generatedAt": "2026-10-12",
    "license": "CC-BY-4.0",
    "method": "https://www.anchorterminal.com/benchmark/",
    "methodology": "0.4",
    "openapi": "https://www.anchorterminal.com/openapi.json",
    "preview": false,
    "run": "2026-10-01",
    "runLabel": "October 2026 research run"
  }
}