{
  "fixes": {
    "slug": "agentos",
    "name": "AgentOS",
    "listing": "https://www.anchorterminal.com/tools/agentos",
    "markdown": "# Fix list: AgentOS\n\nFrom Anchor Terminal's listing at https://www.anchorterminal.com/tools/agentos, the October 2026 research run, assessed 8 October 2026. Grade BB, 75.3 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 AgentOS: 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). Self-hosted rule. The package is free under Apache 2.0 with nothing to buy (20), installs with no card (20) and needs no account (20). No paid tier of AgentOS was found.\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. Reliability, 77 out of 100, up to 4.6 more on the total\n\nWhy it scored 77: Official npm package with Node 22 or newer stated (20). A public CI workflow builds, lints, type-checks and runs the tests with coverage, and the 12 most recent runs on master, all on 8 October 2026, passed (25). 16 open issues and pull requests in total, most filed by the maintainer, with one outside suggestion from 12 September unanswered (20). Written release rules give a breaking change a minor version and a changelog note, but feature commits ship as patches and two breaking minors, 0.11.0 and 0.12.0, landed on 7 October (12). Version 0.12.12, not declared stable (0). Local-software reading.\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## 3. Agent ergonomics, 74 out of 100, up to 4.2 more on the total\n\nWhy it scored 74: An agent is about five lines with the provider detected from the environment, and capability discovery selects tools by meaning, but no MCP client was found in the package source or the docs sitemap, which this line is read on for a framework (12). Session history is capped at about 120K tokens by default, with `history.maxTokens`, `reseed()` and `maxSteps` (20). Typed error classes and a `fallbacksUsed` field on results, with error documentation spread across pages (14). Checkpoints replay or re-run a node by its declared effect class on resume, and providers fall back in order, though the shipped checkpoint store is in memory (18). Few required options, TypeScript only (10).\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## 4. Security \u0026 auth, 81 out of 100, up to 3.3 more on the total\n\nWhy it scored 81: No analytics or telemetry code found in the source, and the privacy and security pages say the runtime sends nothing to the vendor. OpenTelemetry is opt-in and the host starts the SDK (30). Approval gates on five triggers, five security tiers and a rule that composed tools chain a side-effecting step only when the host lists it. Agent-written code runs in an in-process node:vm context, off by default, and no gate is on unless configured (17). SECURITY.md names model output, tool results and retrieved content as untrusted input, and the injection classifier is a separate extension package (11). OpenTelemetry spans, a usage ledger and a reasoning trace (15). SECURITY.md with a private advisory form, acknowledgement in 5 business days and disclosure at 90 days. No security.txt, no bounty and no published advisory found, and the site's security page gives a different acknowledgement time of 48 hours (8). Framework reading.\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## 5. Transparency \u0026 trust, 64 out of 100, up to 3.2 more on the total\n\nMade of editorial 80, provenance 48.\n\nWhy it scored 64: Apache 2.0 (30). The privacy page and the security page agree that the core collects nothing and that the operator is the data controller. Both describe a self-hosted package, and the entity is written three ways, as Framers Lab, Inc., Frame.dev and Framers (22). SECURITY.md says only the latest version gets fixes, and breaking changes carry changelog notes, but no deprecation policy or notice period was found (8). No telemetry is sent and the site says so (20).\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: agentos.sh, registered 2025-10-26 (under a year) (0 of 15)\n- Terms of service: read, states 2 of the 7 things a reader expects (5.7 of 10)\n- Privacy policy: read, states 2 of the 8 things a reader expects (5.5 of 10)\n- Status page: not found (0 of 10)\n- security.txt: not found (0 of 10)\n\n## 6. Schema \u0026 documentation, 92 out of 100, up to 1.3 more on the total\n\nWhy it scored 92: Typed TypeScript API with a TypeDoc reference (25). llms.txt on agentos.sh and docs.agentos.sh (10). The guides say which surface to reach for, for example agency-level approval for most apps and workflow nodes for an explicit graph, and the tool-forging page states what its sandbox does not do (16). Tools take Zod schemas (15). 22 runnable examples and a cookbook, with CLI error codes documented but no single page for the library's error classes found (11). A dated changelog generated on every release and written version rules (15). Framework reading.\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, 87 out of 100, up to 1.1 more on the total\n\nWhy it scored 87: 0.12.12 on 2026-10-08 (30). 98 versions on npm in the 90 days to 8 October (20). Little outside traffic to judge by. Two outside issues were closed with five comments each, one outside suggestion has been open without a reply since 12 September, and nearly all pull requests come from one maintainer (15). The npm package is current (15). CI and coverage run on every push, but the workflows have no `npm audit` step and the repository has no Dependabot configuration file (7).\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- 2026-10-08. The README, which is also the npm page, says forged tools run in a \"hardened `node:vm` sandbox\". The project's own CI fails the build if its architecture docs use that word for the same executor, with the message that the documentation claims an isolation the executor does not have, and the docs quote Node's statement that node:vm is not a security mechanism. The site's security page also says npm audit runs in CI, and no workflow runs it. Sandbox mode is off by default and the docs are candid, so 2 points. https://github.com/framerslab/agentos/blob/master/.github/workflows/ci.yml\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- The scout's facts held on 8 October 2026 (677 stars, 0.12.12, Apache 2.0). Its entry did not mention that no MCP client was found\n- Whether an MCP client exists in a sibling package. None was found in @framers/agentos, in the dependencies of @framers/agentos-extensions 1.0.3 or in the docs sitemap, and we did not clone the extensions repository\n- The npm time index lists versions 1.0.0 to 1.0.3 from 11 December 2025 that are no longer in the registry. We found no note explaining their removal\n- Whether Dependabot is enabled in the repository settings, as the security page says. No configuration file is in the repository\n- The site footer gives MIT for extensions, and @framers/agentos-extensions 1.0.3 declares Apache-2.0 on npm\n- unchecked: the vendor's LongMemEval figures (85.6 per cent on S, 70.2 per cent on M). They are the vendor's claims and were not reproduced\n- unchecked: frame.dev and the hosted Voice Chat Assistant terms, which cover a separate product\n- Which security tier, if any, applies to a plain `agent()` call. The `tier` field is optional and we did not trace the default\n\n## Weaknesses\n\n- Version 0.12.12, with 27 releases on 7 and 8 October 2026 and breaking minors 0.11.0 and 0.12.0 on the same day\n- No MCP client found in the package source or the docs sitemap\n- Agent-written code tools run in an in-process node:vm context, which Node says is not a security mechanism. The mode is off by default\n- The README calls that sandbox hardened, a word the project's CI bars from its architecture docs\n- The only checkpoint store shipped is in memory, so resuming after a crash needs a store you write\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- Pin an exact version of @framers/agentos. Feature commits ship as patch releases while it is 0.x, and breaking changes ship as minors\n- Leave emergentConfig.allowSandboxTools off unless forged code runs through an executor that isolates. The default executor shares the host process\n- Pass your own ICheckpointStore to compile() for runs that must survive a restart. The default store is in memory\n- Set hitl.approvals.beforeTool for tools that write, and choose a handler. No approval gate is on by default\n- Use Node 22 or newer, and set a provider key such as OPENAI_API_KEY or ANTHROPIC_API_KEY. The provider is detected from the environment\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": "BB",
    "score": 75.3,
    "assessed": "2026-10-08",
    "run": "October 2026 research run",
    "categories": [
      {
        "key": "payments",
        "name": "Payments \u0026 pricing",
        "score": 60,
        "maxGain": 5,
        "reason": "No payment protocol (0). Self-hosted rule. The package is free under Apache 2.0 with nothing to buy (20), installs with no card (20) and needs no account (20). No paid tier of AgentOS was found.",
        "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": "reliability",
        "name": "Reliability",
        "score": 77,
        "maxGain": 4.6,
        "reason": "Official npm package with Node 22 or newer stated (20). A public CI workflow builds, lints, type-checks and runs the tests with coverage, and the 12 most recent runs on master, all on 8 October 2026, passed (25). 16 open issues and pull requests in total, most filed by the maintainer, with one outside suggestion from 12 September unanswered (20). Written release rules give a breaking change a minor version and a changelog note, but feature commits ship as patches and two breaking minors, 0.11.0 and 0.12.0, landed on 7 October (12). Version 0.12.12, not declared stable (0). Local-software reading.",
        "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": "ergonomics",
        "name": "Agent ergonomics",
        "score": 74,
        "maxGain": 4.2,
        "reason": "An agent is about five lines with the provider detected from the environment, and capability discovery selects tools by meaning, but no MCP client was found in the package source or the docs sitemap, which this line is read on for a framework (12). Session history is capped at about 120K tokens by default, with `history.maxTokens`, `reseed()` and `maxSteps` (20). Typed error classes and a `fallbacksUsed` field on results, with error documentation spread across pages (14). Checkpoints replay or re-run a node by its declared effect class on resume, and providers fall back in order, though the shipped checkpoint store is in memory (18). Few required options, TypeScript only (10).",
        "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": 81,
        "maxGain": 3.3,
        "reason": "No analytics or telemetry code found in the source, and the privacy and security pages say the runtime sends nothing to the vendor. OpenTelemetry is opt-in and the host starts the SDK (30). Approval gates on five triggers, five security tiers and a rule that composed tools chain a side-effecting step only when the host lists it. Agent-written code runs in an in-process node:vm context, off by default, and no gate is on unless configured (17). SECURITY.md names model output, tool results and retrieved content as untrusted input, and the injection classifier is a separate extension package (11). OpenTelemetry spans, a usage ledger and a reasoning trace (15). SECURITY.md with a private advisory form, acknowledgement in 5 business days and disclosure at 90 days. No security.txt, no bounty and no published advisory found, and the site's security page gives a different acknowledgement time of 48 hours (8). Framework reading.",
        "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": "transparency",
        "name": "Transparency \u0026 trust",
        "score": 64,
        "maxGain": 3.2,
        "reason": "Apache 2.0 (30). The privacy page and the security page agree that the core collects nothing and that the operator is the data controller. Both describe a self-hosted package, and the entity is written three ways, as Framers Lab, Inc., Frame.dev and Framers (22). SECURITY.md says only the latest version gets fixes, and breaking changes carry changelog notes, but no deprecation policy or notice period was found (8). No telemetry is sent and the site says so (20).",
        "blend": "editorial 80, provenance 48",
        "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": 92,
        "maxGain": 1.3,
        "reason": "Typed TypeScript API with a TypeDoc reference (25). llms.txt on agentos.sh and docs.agentos.sh (10). The guides say which surface to reach for, for example agency-level approval for most apps and workflow nodes for an explicit graph, and the tool-forging page states what its sandbox does not do (16). Tools take Zod schemas (15). 22 runnable examples and a cookbook, with CLI error codes documented but no single page for the library's error classes found (11). A dated changelog generated on every release and written version rules (15). Framework reading.",
        "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": 87,
        "maxGain": 1.1,
        "reason": "0.12.12 on 2026-10-08 (30). 98 versions on npm in the 90 days to 8 October (20). Little outside traffic to judge by. Two outside issues were closed with five comments each, one outside suggestion has been open without a reply since 12 September, and nearly all pull requests come from one maintainer (15). The npm package is current (15). CI and coverage run on every push, but the workflows have no `npm audit` step and the repository has no Dependabot configuration file (7).",
        "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": "agentos.sh, registered 2025-10-26 (under a year)",
        "points": 0,
        "max": 15
      },
      {
        "label": "Terms of service",
        "value": "read, states 2 of the 7 things a reader expects",
        "points": 5.7,
        "max": 10
      },
      {
        "label": "Privacy policy",
        "value": "read, states 2 of the 8 things a reader expects",
        "points": 5.5,
        "max": 10
      },
      {
        "label": "Status page",
        "value": "not found",
        "points": 0,
        "max": 10
      },
      {
        "label": "security.txt",
        "value": "not found",
        "points": 0,
        "max": 10
      }
    ],
    "deductions": [
      "2026-10-08. The README, which is also the npm page, says forged tools run in a \"hardened `node:vm` sandbox\". The project's own CI fails the build if its architecture docs use that word for the same executor, with the message that the documentation claims an isolation the executor does not have, and the docs quote Node's statement that node:vm is not a security mechanism. The site's security page also says npm audit runs in CI, and no workflow runs it. Sandbox mode is off by default and the docs are candid, so 2 points. https://github.com/framerslab/agentos/blob/master/.github/workflows/ci.yml"
    ],
    "unchecked": [
      "The scout's facts held on 8 October 2026 (677 stars, 0.12.12, Apache 2.0). Its entry did not mention that no MCP client was found",
      "Whether an MCP client exists in a sibling package. None was found in @framers/agentos, in the dependencies of @framers/agentos-extensions 1.0.3 or in the docs sitemap, and we did not clone the extensions repository",
      "The npm time index lists versions 1.0.0 to 1.0.3 from 11 December 2025 that are no longer in the registry. We found no note explaining their removal",
      "Whether Dependabot is enabled in the repository settings, as the security page says. No configuration file is in the repository",
      "The site footer gives MIT for extensions, and @framers/agentos-extensions 1.0.3 declares Apache-2.0 on npm",
      "unchecked: the vendor's LongMemEval figures (85.6 per cent on S, 70.2 per cent on M). They are the vendor's claims and were not reproduced",
      "unchecked: frame.dev and the hosted Voice Chat Assistant terms, which cover a separate product",
      "Which security tier, if any, applies to a plain `agent()` call. The `tier` field is optional and we did not trace the default"
    ],
    "weaknesses": [
      "Version 0.12.12, with 27 releases on 7 and 8 October 2026 and breaking minors 0.11.0 and 0.12.0 on the same day",
      "No MCP client found in the package source or the docs sitemap",
      "Agent-written code tools run in an in-process node:vm context, which Node says is not a security mechanism. The mode is off by default",
      "The README calls that sandbox hardened, a word the project's CI bars from its architecture docs",
      "The only checkpoint store shipped is in memory, so resuming after a crash needs a store you write"
    ],
    "agentNotes": [
      "Pin an exact version of @framers/agentos. Feature commits ship as patch releases while it is 0.x, and breaking changes ship as minors",
      "Leave emergentConfig.allowSandboxTools off unless forged code runs through an executor that isolates. The default executor shares the host process",
      "Pass your own ICheckpointStore to compile() for runs that must survive a restart. The default store is in memory",
      "Set hitl.approvals.beforeTool for tools that write, and choose a handler. No approval gate is on by default",
      "Use Node 22 or newer, and set a provider key such as OPENAI_API_KEY or ANTHROPIC_API_KEY. The provider is detected from the environment"
    ],
    "recheck": "https://www.anchorterminal.com/builders/#disputes"
  },
  "meta": {
    "attribution": "Anchor Terminal (https://www.anchorterminal.com)",
    "docs": "https://www.anchorterminal.com/docs/",
    "generatedAt": "2026-10-08",
    "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"
  }
}