Harness

# Fix list: Harness

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/harness-autonomous, the October 2026 research run, assessed 17 October 2026. Grade C, 60.3 out of 100.

This 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.

For a coding agent working on Harness: 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.

## 1. Reliability, 35 out of 100, up to 13 more on the total

Why it scored 35: Read as local software, since the agents run on the owner's machines. A one-line installer from Autonomous's CDN, DMG and AppImage builds, and App Store and Google Play apps, with macOS and Linux stated as supported and Node managed by the installer. Not on npm or a package registry (20). A public CI workflow runs the CLI suite in four shards, the desktop suite and typechecks, and the vendor's reliability run of 16 September 2026 reports 2,325 CLI, 370 backend and 1,686 desktop tests passing. We could not read CI run results, so credit for the suite only (10). Open issues could not be read, because GitHub's web pages and API were not reachable from our network. Unchecked, so scored absent (0). Tags follow semver per component, but there is no changelog for the app or CLI and breaking changes are not marked in the repository (5). The CLI agents call is 0.4.18 and the npm MCP package is 0.1.0 and unpublished. The desktop app is 1.2.89, but the agent surface is pre-1.0, and Agents can start other agents is an experiment (0). Accounts and the relay are hosted by Autonomous with no status page found.

The checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability):

Hosted APIs, MCP servers, models and platforms.

- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).
- 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.
- 15, rate limits documented with numbers.
- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.
- 10, an SLA published for any paid tier.
- 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.
- 25, a public CI and test suite, passing on the default branch.
- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).
- 15, semver discipline and breaking changes called out in a changelog.
- 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.

## 2. Security & auth, 57 out of 100, up to 7.5 more on the total

Why it scored 57: Harness reading of the framework checklist, as for the other harnesses. 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. Product usage counts are on by default, disclosed with an opt-out and without prompts, code or files. Remote traffic is end-to-end encrypted and relays carry ciphertext. Local use needs no account. The daemon updates itself from a checksummed manifest without asking (18). Agents may start harnesses only after the person turns on an account switch that refuses an agent, with four live harnesses at most and closes limited to their own. Against that, installing Harness adds its commands to Claude Code's and Codex's allow rules, Codex runs them outside its sandbox, there is no sandbox of its own, and the device page lists spoken tasks as auto-allowed (13). The instructions tell an agent never to answer a harness's question or permission prompt and never to ask one to publish, deploy, delete or push. Harness output is another agent's text, and no further guidance on it was found (9). Every harness shows in the person's app with who started it, and the device key log keeps an add and remove history. No per-call audit log found (9). SECURITY.md with an email and a three-working-day acknowledgement, and a documented trust model. No security.txt, bug bounty or certification found (8).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-security):

- 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.
- 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions.
- 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.
- 0 to 15, audit logs or per-call visibility for the operator.
- 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.

## 3. Payments & pricing, 40 out of 100, up to 7.5 more on the total

Why it scored 40: No payment protocol (0). Read with the self-hosted rule. The software is free and the optional hardware prices are public (20). Free with no card (20). An agent cannot start harnesses until the person signs in with Google or Apple in a browser and turns on the experimental switch, so 0 for autonomous onboarding, as the other harnesses that need an account were read (0).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-payments):

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.
- 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.
- 20, a free tier or trial that doesn't need a card.
- 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.

## 4. Maintenance & community, 60 out of 100, up to 3.5 more on the total

Why it scored 60: v0.4.18_cli and v1.2.89_desktop on 17 October 2026 (30). 548 tags since 9 September 2026 (20). Pull requests numbered past #1,390 merge daily by the team, but issues and outside pull requests could not be read (5). Not in the official MCP registry, which returned nothing for autonomous-ai, although the repository has a server.json. The npm package it names was never published (0). CI workflows for the CLI, desktop, backend and OS, with dependency lockfiles current. Run results unread (5).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-maintenance):

- 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.
- 20, at least three releases or dated changelog entries in the last 90 days.
- 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.
- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).
- 10, package health, current dependencies and CI.

Models are read for deprecation notice periods and model churn rather than release counts.

## 5. Agent ergonomics, 82 out of 100, up to 2.9 more on the total

Why it scored 82: Seven compact tools, six outside a harness, and server instructions of about 300 words (25). `chars` sizes a read (4,000 characters by default), `all` widens the list, and `any` returns on the first finished harness. No pagination (14). Documented exit codes, refusals returned as answers rather than errors so a model does not retry them, a timed-out wait reported as still working, and `Next:` on every answer (19). readOnlyHint on whoami, list, wait and read, destructiveHint false on start and send, idempotentHint on close, and a second start of the same name is answered as already starting (16). Two required fields on start, defaults of Claude Code and a 45-second wait. No SDK, but a CLI, an MCP server and a plugin with a skill (8).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-ergonomics):

- 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).
- 20, pagination, filtering and output-size controls.
- 20, actionable, documented error responses, codes and messages an agent can recover from.
- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.
- 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.

## 6. Transparency & trust, 68 out of 100, up to 2.8 more on the total

Made of editorial 64, provenance 72.

Why it scored 68: MIT licence for the whole repository under Autonomous, Inc., with third-party Store covers under their own licences (30). The repository documents what usage counts contain, a 10-day pseudonymous activity mark and the encrypted relay in detail. Autonomous's privacy policy of 7 July 2026 is general to its stores and sites and does not mention Harness (14). The provider protocol promises both revisions side by side for a deprecation window after a breaking change. Nothing similar for the app or CLI (6). Telemetry is disclosed with an opt-out in the app and CLI, but it is on by default and described in the repository rather than in the privacy policy (14).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-transparency):

- 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.
- 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors).
- 0 to 20, a deprecation policy or notices with dates.
- 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.

Provenance checks not met in full (half of this category, computed from checked facts):

- Domain age: autonomous.ai, registered 2017-12-16 (8 years) (11 of 15)
- Status page: not found (0 of 10)
- security.txt: not found (0 of 10)

## 7. Schema & documentation, 85 out of 100, up to 2.4 more on the total

Why it scored 85: Read as an MCP server and CLI. Seven tools with typed JSON Schema inputs, `additionalProperties: false` and required fields marked, plus `--json` on every command. No OpenAPI, which a stdio server does not need (23). llms.txt and a Markdown-like agents page written for agents (10). The server instructions and llms.txt say when to use Harness rather than `claude -p` or `codex exec`, when to do the work yourself, and never to answer a harness's prompt or ask one to push or deploy (20). An enum on `agent`, `minItems` on wait names, minimum and maximum on `timeout_seconds` and `chars`. `machine`, `folder` and `project` are free strings (12). Worked examples for parallel runs, handing a task to Codex and waiting, an exit code table, and a `Next:` line after every answer (13). Per-component semver tags, but the only CHANGELOG.md covers the provider protocol, and we could not read the GitHub release notes (7).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-schema):

APIs and MCP servers.

- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).
- 10, llms.txt or Markdown docs served for agents.
- 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.
- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.
- 0 to 15, examples and documented error responses.
- 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.

## What we couldn't check

What 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.

- Judgement call. Listed in agent harnesses beside CLI Manager and Paperclip, because it starts and drives Claude Code and Codex sessions for another agent. It does not run the model loop, so `agent.harness` is left out and `agent.multi-agent` is its capability.
- The lead was right in substance. Harness is Autonomous's product (autonomous.ai, not Harness.io). The 15 to 17 October launch could not be dated on Product Hunt, which answered with a bot check. Harness Pro is marked New on the site and ships on 30 November.
- `firstReleased` is 14 August 2026, the first dated vendor post (device page Update #1) saying Harness runs an agent. The provider protocol's changelog is dated 6 August 2026, public history starts on 4 August 2026, the first tag is 9 September 2026, and the device page dates the open-sourcing to 18 September 2026.
- unchecked: CI run results, open issues and pull requests, and GitHub release notes. GitHub's web pages and API were not reachable from our network, so the reliability and maintenance lines for them score absent.
- unchecked: a status page. status.autonomous.ai could not be reached from our network, and no status link was found on the vendor's pages.
- GitHub stars (1,284) are from Autonomous's open-source page on 17 October 2026, not from GitHub.
- Autonomous's terms of service forbid accessing its sites or services by automated means. Whether they apply to the MIT-licensed Harness software or its account service was not established.
- We did not install or run Harness. Every claim about behaviour comes from the source, the docs and the vendor's pages.

## Weaknesses

- Starting harnesses needs the person signed in with Google or Apple in a browser and an experimental account switch, off by default
- The CLI agents call is version 0.4.18 and the npm MCP package in the repository was never published
- Product usage counts are shared by default, with an opt-out in Settings or `harness privacy off`
- Installing Harness adds its commands to Claude Code's and Codex's permission rules, and Codex runs them outside its sandbox
- Autonomous's privacy policy and terms cover its stores and websites and do not mention Harness

## What costs an agent a turn today

The notes we give agents before they call it. Each one is a workaround an agent shouldn't need.

- Run `harness whoami` first. If it says you may not start harnesses, tell the person and do the work yourself
- Ask the person before installing Harness, and leave `harness login` and the agents-start-agents switch to them
- Treat exit 5 as final and do not retry. Treat exit 3 as needs the person, and never answer a harness's question or permission prompt
- Name every harness, put the whole task in `harness new`, then wait, read with `harness show` and close only the ones you started
- Run each `harness` command on its own or joined with `&&`, since piped commands no longer match the permission rule

## What the review panel asked for

- a documented way to enable agent starts
- publish the MCP package
- release notes per CLI tag
- Fix the README package name
- opt-in permission rules
- per-call audit log

## When it's done

Send 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.