{
  "fixes": {
    "slug": "strands-decider",
    "name": "Strands Decider 2B",
    "listing": "https://www.anchorterminal.com/tools/strands-decider",
    "markdown": "# Fix list: Strands Decider 2B\n\nFrom Anchor Terminal's listing at https://www.anchorterminal.com/tools/strands-decider, the October 2026 research run, assessed 5 October 2026. Grade C, 61.3 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 Strands Decider 2B: 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. Reliability, 50 out of 100, up to 10 more on the total\n\nWhy it scored 50: Scored on the local-package checklist, since Strands Decider is open weights the owner runs, as for Kev and Laya. `pip install strands-decider` from PyPI, 0.1.0 of 1 October 2026, with Python 3.10 or later stated and CUDA, MPS, MLX and CPU extras (20). A public suite of 29 test files runs on GitHub Actions on Python 3.10 with the floor pins and 3.12 with the newest, but GitHub's web pages and API were blocked for our reader, so we couldn't see whether main is passing (15 of 25). Open crash and regression issues couldn't be read for the same reason. Commits reference pull requests up to #35 in five days (10 of 25, unchecked). No changelog file and no GitHub release tags. Commit titles follow a conventional format checked in CI, and each model version is a separate Hugging Face repository with its results (5 of 15). 0.1.0, and the package calls itself experimental (0).\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## 2. Security \u0026 auth, 49 out of 100, up to 8.9 more on the total\n\nWhy it scored 49: Read as software you run. No account. The server binds to 127.0.0.1 and has no authentication option, and the README says to use it for local experiments (8 of 30). A decision model has no write actions, so there's nothing to approve (15 of 20). The model card warns that questions are read less than documents and that calibration holds only on short classification. Nothing documents how hostile text in the state can steer an answer, though the launch example uses it as a tool-call guardrail (6 of 15). Each response carries token usage and latency, and `/health` reports the checkpoint and device. No request log (5 of 15). SECURITY.md routes reports to the AWS Vulnerability Disclosure Program on HackerOne. A weekly pip-audit, dependency review on pull requests, actions pinned by commit and trusted publishing to PyPI. The head ships as safetensors with a SHA-256 manifest and a `verify` command. No security.txt on strandsagents.com (15 of 20).\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## 3. Agent ergonomics, 69 out of 100, up to 5 more on the total\n\nWhy it scored 69: Read as an API an agent calls for a decision, as for the other decision models. Answers are a probability per option, and the state is read once with each extra question adding only its own tokens. The window is 4,096 tokens, and by default an over-long state is cut to fit without an error (15 of 25). The caller sets the questions, any number a request, with `--max-batch` (default 32) setting how many go in one forward pass. No batch-of-states route (15 of 20). Validation and engine errors return 422 with the message, and `--strict-window` names the window. Not documented as a table (13 of 20). Calls are stateless and safe to retry. No retry guidance, and the server runs one worker with no rate limiting (15 of 20). One pip install with a CLI and a Python class, and `strands-decider-latest` as the default model. Python only, and the code says Jev compatibility isn't verified (11 of 15).\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. Payments \u0026 pricing, 60 out of 100, up to 5 more on the total\n\nWhy it scored 60: Free Apache-2.0 software with nothing to buy, so 20 + 20 + 20 for pricing, free use and no sign-up, by the self-hosted rule. No payment protocol (0). No hosted API was found, on Amazon Bedrock or elsewhere, so there's no hosted option to grade.\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## 5. Transparency \u0026 trust, 54 out of 100, up to 4 more on the total\n\nMade of editorial 63, provenance 44.\n\nWhy it scored 54: Apache-2.0 for the code and checkpoints, on an Apache-2.0 base, with the training recipe, data inventory, configs, data hashes and evaluation logs published (30). Self-hosted, so inputs stay on the operator's hardware by construction, but we found no statement saying so. The licence notes list training sources whose Hub licence tags include other, unknown and none (15 of 30). v19 stays published after v21 replaced it as the reference. No deprecation policy (8 of 20). No telemetry code found in the package and no statement either way. Weights download through the Hugging Face Hub client (10 of 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: strandsagents.com, registered 2025-05-15 (1 year) (3 of 15)\n- Status page: not found (0 of 10)\n- Changelog: not found (0 of 10)\n- security.txt: not found (0 of 10)\n\n## 6. Schema \u0026 documentation, 76 out of 100, up to 3.9 more on the total\n\nWhy it scored 76: Read for a model you serve yourself. The server is FastAPI with Pydantic request and response models (`SystemOneRequest`, `NoulQuestion`, `ChoiceQuestion`, `ScoreQuestion`) following TypeSafe's public Jev documentation, with no spec file of its own published (18 of 25). strandsagents.com's llms.txt lists the launch post with a Markdown twin, and the repository docs are Markdown. No llms.txt entry for the model's docs (7 of 10). The README, launch post and model card say what it's for and where it fails, with measured figures for each limitation in evaluation/README.md (18 of 20). Three question types, 2 to 255 options a choice, 2 to 10 levels a score, `noul` criteria keys limited to true and false, and at least one question. State is free-form by design (12 of 15). CLI, curl and Python examples with sample output. Caller errors return 422 with a message, but there's no error table (11 of 15). Model versions are named repositories (v19, v21) and research/README.md lists every training run, with no package changelog (10 of 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, 79 out of 100, up to 1.8 more on the total\n\nWhy it scored 79: Read for an open-weight model. v21 weights on 5 October 2026 and the 0.1.0 package on 1 October (30). Three releases in the window, the v19 weights (repository created 30 September), the 0.1.0 package (1 October) and the v21 weights (5 October) (20). 11 commits from 6 authors between 1 and 5 October, with pull requests merged daily and a Discord channel, but issue reply times couldn't be read (12 of 25, unchecked). A Python package and CLI from the vendor, with a Strands agent example and integration libraries promised without a date. Not an MCP server (10 of 15). CI with pinned actions and a weekly dependency audit. Pass state unchecked, and the MLX extra isn't in a release yet (7 of 10).\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## 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- unchecked: whether CI passes on main. GitHub's web pages and API were blocked for our reader, though the repository cloned\n- unchecked: open issues, pull requests and reply times, for the same reason\n- unchecked: GitHub stars and forks, so popularity is blank. Hugging Face showed 57 likes and 0 downloads for v19\n- No hosted API was found. The launch post and repository mention Amazon Bedrock only as the LLM in the agent example and as a data-generation backend\n- The accuracy and calibration figures, and the 3rd-of-33 board position the launch post cites, are the vendor's. We haven't run them, and the training mix includes BANKING77 and other public datasets that benchmarks in this category use\n- The legal entity is inferred from SECURITY.md and AWS's Strands Labs post. The licence names no copyright holder\n\n## Weaknesses\n\n- Version 0.1.0, described as experimental in its package metadata, with no changelog file\n- A 4,096-token window, and by default an over-long state is cut to fit without an error\n- The local server has no authentication option\n- No hosted API, so the operator runs and scales the model\n- The model card says its calibration is established on short classification only\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 the checkpoint by its full name, such as `StrandsAgents/strands-decider-2B-hobson-v21`, since each version is a separate Hugging Face repository\n- Start the server with `--strict-window` when a cut state would make an answer wrong. It then returns 422 naming the window\n- Ask every question about one state in one request. The state is read once and each question adds only its own tokens\n- Keep the server on 127.0.0.1 or put an authenticating proxy in front. It has no key option\n- Measure thresholds on your own traffic before acting automatically. The card says confidence bands hold for short classification only\n\n## What the review panel asked for\n\n- a dated changelog\n- a checkpoint retention policy\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": "C",
    "score": 61.3,
    "assessed": "2026-10-05",
    "run": "October 2026 research run",
    "categories": [
      {
        "key": "reliability",
        "name": "Reliability",
        "score": 50,
        "maxGain": 10,
        "reason": "Scored on the local-package checklist, since Strands Decider is open weights the owner runs, as for Kev and Laya. `pip install strands-decider` from PyPI, 0.1.0 of 1 October 2026, with Python 3.10 or later stated and CUDA, MPS, MLX and CPU extras (20). A public suite of 29 test files runs on GitHub Actions on Python 3.10 with the floor pins and 3.12 with the newest, but GitHub's web pages and API were blocked for our reader, so we couldn't see whether main is passing (15 of 25). Open crash and regression issues couldn't be read for the same reason. Commits reference pull requests up to #35 in five days (10 of 25, unchecked). No changelog file and no GitHub release tags. Commit titles follow a conventional format checked in CI, and each model version is a separate Hugging Face repository with its results (5 of 15). 0.1.0, and the package calls itself experimental (0).",
        "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": "security",
        "name": "Security \u0026 auth",
        "score": 49,
        "maxGain": 8.9,
        "reason": "Read as software you run. No account. The server binds to 127.0.0.1 and has no authentication option, and the README says to use it for local experiments (8 of 30). A decision model has no write actions, so there's nothing to approve (15 of 20). The model card warns that questions are read less than documents and that calibration holds only on short classification. Nothing documents how hostile text in the state can steer an answer, though the launch example uses it as a tool-call guardrail (6 of 15). Each response carries token usage and latency, and `/health` reports the checkpoint and device. No request log (5 of 15). SECURITY.md routes reports to the AWS Vulnerability Disclosure Program on HackerOne. A weekly pip-audit, dependency review on pull requests, actions pinned by commit and trusted publishing to PyPI. The head ships as safetensors with a SHA-256 manifest and a `verify` command. No security.txt on strandsagents.com (15 of 20).",
        "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": "ergonomics",
        "name": "Agent ergonomics",
        "score": 69,
        "maxGain": 5,
        "reason": "Read as an API an agent calls for a decision, as for the other decision models. Answers are a probability per option, and the state is read once with each extra question adding only its own tokens. The window is 4,096 tokens, and by default an over-long state is cut to fit without an error (15 of 25). The caller sets the questions, any number a request, with `--max-batch` (default 32) setting how many go in one forward pass. No batch-of-states route (15 of 20). Validation and engine errors return 422 with the message, and `--strict-window` names the window. Not documented as a table (13 of 20). Calls are stateless and safe to retry. No retry guidance, and the server runs one worker with no rate limiting (15 of 20). One pip install with a CLI and a Python class, and `strands-decider-latest` as the default model. Python only, and the code says Jev compatibility isn't verified (11 of 15).",
        "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": "payments",
        "name": "Payments \u0026 pricing",
        "score": 60,
        "maxGain": 5,
        "reason": "Free Apache-2.0 software with nothing to buy, so 20 + 20 + 20 for pricing, free use and no sign-up, by the self-hosted rule. No payment protocol (0). No hosted API was found, on Amazon Bedrock or elsewhere, so there's no hosted option to grade.",
        "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": "transparency",
        "name": "Transparency \u0026 trust",
        "score": 54,
        "maxGain": 4,
        "reason": "Apache-2.0 for the code and checkpoints, on an Apache-2.0 base, with the training recipe, data inventory, configs, data hashes and evaluation logs published (30). Self-hosted, so inputs stay on the operator's hardware by construction, but we found no statement saying so. The licence notes list training sources whose Hub licence tags include other, unknown and none (15 of 30). v19 stays published after v21 replaced it as the reference. No deprecation policy (8 of 20). No telemetry code found in the package and no statement either way. Weights download through the Hugging Face Hub client (10 of 20).",
        "blend": "editorial 63, provenance 44",
        "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": 76,
        "maxGain": 3.9,
        "reason": "Read for a model you serve yourself. The server is FastAPI with Pydantic request and response models (`SystemOneRequest`, `NoulQuestion`, `ChoiceQuestion`, `ScoreQuestion`) following TypeSafe's public Jev documentation, with no spec file of its own published (18 of 25). strandsagents.com's llms.txt lists the launch post with a Markdown twin, and the repository docs are Markdown. No llms.txt entry for the model's docs (7 of 10). The README, launch post and model card say what it's for and where it fails, with measured figures for each limitation in evaluation/README.md (18 of 20). Three question types, 2 to 255 options a choice, 2 to 10 levels a score, `noul` criteria keys limited to true and false, and at least one question. State is free-form by design (12 of 15). CLI, curl and Python examples with sample output. Caller errors return 422 with a message, but there's no error table (11 of 15). Model versions are named repositories (v19, v21) and research/README.md lists every training run, with no package changelog (10 of 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": 79,
        "maxGain": 1.8,
        "reason": "Read for an open-weight model. v21 weights on 5 October 2026 and the 0.1.0 package on 1 October (30). Three releases in the window, the v19 weights (repository created 30 September), the 0.1.0 package (1 October) and the v21 weights (5 October) (20). 11 commits from 6 authors between 1 and 5 October, with pull requests merged daily and a Discord channel, but issue reply times couldn't be read (12 of 25, unchecked). A Python package and CLI from the vendor, with a Strands agent example and integration libraries promised without a date. Not an MCP server (10 of 15). CI with pinned actions and a weekly dependency audit. Pass state unchecked, and the MLX extra isn't in a release yet (7 of 10).",
        "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": "strandsagents.com, registered 2025-05-15 (1 year)",
        "points": 3,
        "max": 15
      },
      {
        "label": "Status page",
        "value": "not found",
        "points": 0,
        "max": 10
      },
      {
        "label": "Changelog",
        "value": "not found",
        "points": 0,
        "max": 10
      },
      {
        "label": "security.txt",
        "value": "not found",
        "points": 0,
        "max": 10
      }
    ],
    "unchecked": [
      "unchecked: whether CI passes on main. GitHub's web pages and API were blocked for our reader, though the repository cloned",
      "unchecked: open issues, pull requests and reply times, for the same reason",
      "unchecked: GitHub stars and forks, so popularity is blank. Hugging Face showed 57 likes and 0 downloads for v19",
      "No hosted API was found. The launch post and repository mention Amazon Bedrock only as the LLM in the agent example and as a data-generation backend",
      "The accuracy and calibration figures, and the 3rd-of-33 board position the launch post cites, are the vendor's. We haven't run them, and the training mix includes BANKING77 and other public datasets that benchmarks in this category use",
      "The legal entity is inferred from SECURITY.md and AWS's Strands Labs post. The licence names no copyright holder"
    ],
    "weaknesses": [
      "Version 0.1.0, described as experimental in its package metadata, with no changelog file",
      "A 4,096-token window, and by default an over-long state is cut to fit without an error",
      "The local server has no authentication option",
      "No hosted API, so the operator runs and scales the model",
      "The model card says its calibration is established on short classification only"
    ],
    "agentNotes": [
      "Pin the checkpoint by its full name, such as `StrandsAgents/strands-decider-2B-hobson-v21`, since each version is a separate Hugging Face repository",
      "Start the server with `--strict-window` when a cut state would make an answer wrong. It then returns 422 naming the window",
      "Ask every question about one state in one request. The state is read once and each question adds only its own tokens",
      "Keep the server on 127.0.0.1 or put an authenticating proxy in front. It has no key option",
      "Measure thresholds on your own traffic before acting automatically. The card says confidence bands hold for short classification only"
    ],
    "requests": [
      {
        "text": "a dated changelog",
        "reviews": 1
      },
      {
        "text": "a checkpoint retention policy",
        "reviews": 1
      }
    ],
    "recheck": "https://www.anchorterminal.com/builders/#disputes"
  },
  "meta": {
    "attribution": "Anchor Terminal (https://www.anchorterminal.com)",
    "docs": "https://www.anchorterminal.com/docs/",
    "generatedAt": "2026-10-06",
    "license": "CC-BY-4.0",
    "method": "https://www.anchorterminal.com/benchmark/",
    "methodology": "0.3",
    "openapi": "https://www.anchorterminal.com/openapi.json",
    "preview": false,
    "run": "2026-10-01",
    "runLabel": "October 2026 research run"
  }
}