# Fix list: AWS MCP Servers From Anchor Terminal's listing at https://www.anchorterminal.com/tools/aws-mcp-servers, the October 2026 research run, assessed 1 October 2026. Grade B, 63.3 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 AWS MCP Servers: 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, 48 out of 100, up to 10.4 more on the total Why it scored 48: The listing covers a hosted server and about 60 local ones, so this is the average of the two rubrics. Hosted Knowledge server (25). The AWS Health Dashboard is a public status page, but it's script-rendered and we didn't confirm a component for this endpoint (10 of 20). No readable incident history for it (5 of 30). The README says it's rate-limited and gives no numbers (0 of 15), documents no 429 or retry behaviour (0 of 15) and no SLA (0). 'This MCP server is in general availability' (10). Local servers (70). Published on PyPI with Python 3.10 to 3.13 stated (20). CI runs builds, tests, CodeQL, Bandit, Semgrep and Scorecard on pull requests, pass state not visible (20 of 25). 200 open issues, with July crash reports for the EC2 and AgentCore servers still marked needs-triage (10 of 25). Breaking changes are marked in commit titles, but per-server changelogs lag. The documentation server's CHANGELOG stops at 1.0.0 while PyPI has 1.2.2 (5 of 15). The documentation (1.2.2) and AWS API (1.5.6) servers are past 1.0 (15). 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. Schema & documentation, 69 out of 100, up to 5 more on the total Why it scored 69: Local servers declare typed inputs through FastMCP and Pydantic. We couldn't read the hosted Knowledge server's own schemas, which aren't in the repository (22 of 25). No llms.txt on awslabs.github.io; the docs are Markdown in the repository (5 of 10). The Knowledge README states each of its 5 tools' purpose and has a FAQ on when to use it instead of the local documentation server, and DESIGN_GUIDELINES.md sets a shared style (12 of 20). Typed parameters with defaults and limits on the documentation tools (11 of 15). README examples. 1.2.2 of the documentation server made read failures raise instead of returning text, flagged as breaking (9 of 15). Date-stamped monorepo releases and per-server changelogs, some of them stale (10 of 15). 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. ## 3. Payments & pricing, 60 out of 100, up to 5 more on the total Why it scored 60: Knowledge is free with no account and no key. That earns pricing published (20), a free tier without a card (20) and keyless access (20). The local servers are free too, and AWS bills the API calls behind them. No x402 (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. Agent ergonomics, 74 out of 100, up to 4.2 more on the total Why it scored 74: Knowledge has 5 tools, and every local server is its own small toolset loaded on its own (23 of 25). Search takes topic filters and a result limit, and reads take a maximum length and start index (16 of 20). Error handling is uneven across 60 servers, with structured `error_type` fields in the billing server and plain exceptions elsewhere (12 of 20). The AWS API server sets readOnlyHint and destructiveHint on `call_aws` from its read-only mode. The documentation server's tools carry no annotations (10 of 20). Knowledge needs no config, local servers start from `uvx` with an AWS profile, and AWS SDKs exist in many languages (13 of 15). 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. ## 5. Security & auth, 84 out of 100, up to 2.8 more on the total Why it scored 84: Knowledge holds no credentials. Local servers use IAM profiles or roles, and the managed ECS and EKS servers take SigV4 through mcp-proxy-for-aws, so access is scoped by IAM policy and revocable (28 of 30). Most servers keep writes off until `--allow-write` and sensitive data off until `--allow-sensitive-data-access`. The AWS API server has `READ_OPERATIONS_ONLY`, `REQUIRE_MUTATION_CONSENT` (elicitation before any non-read call) and a deny and elicit list, but the first two default to false (16 of 20). The AWS API server's README has a prompt-injection section that says not to connect it to untrusted data and to keep a human in the loop. Other servers return logs and records with no such note (10 of 15). Every call made with AWS credentials lands in CloudTrail (14 of 15). Reports go to aws-security@amazon.com, seven advisories were published in public in 2026, and CI runs CodeQL, Scorecard and secret detection. aws.amazon.com's security.txt was past its Expires date per the 26 September check (16 of 20). 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. ## 6. Transparency & trust, 72 out of 100, up to 2.5 more on the total Made of editorial 64, provenance 80. Why it scored 72: The local servers are Apache-2.0. The Knowledge service is closed and runs under the AWS Site Terms (25 of 30). The Knowledge README says telemetry isn't used to train models, but states no retention, and the rest falls under the general AWS Privacy Notice (15 of 30). The Cloud Control API server is marked deprecated with its successor named, and an RFC proposes deprecating the OpenAPI server, with no dates (12 of 20). The AWS API server sends configuration telemetry by default with `AWS_API_MCP_TELEMETRY=false` to stop it. Knowledge says it collects telemetry and mentions no opt-out (12 of 20). 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: api.aws, no registry record we could read (0 of 15) - security.txt: published but past its Expires date (5 of 10) ## 7. Maintenance & community, 79 out of 100, up to 1.8 more on the total Why it scored 79: Release 2026.09.20260930084625 on 30 September, with documentation server 1.2.2 on PyPI the same day (30). 17 dated releases since 3 July (20). Fixes merge most days, but 200 issues are open and July bug reports still carry needs-triage (12 of 25). Only the managed ECS and EKS servers are in the official registry, under aws.api.us-east-1. Knowledge and the local servers aren't, and the Knowledge README names Smithery and Cursor as its registries (7 of 15). Dependency review, Trivy and Scorecard in CI (10). 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. ## Deductions Each comes off the total. A fixed and documented problem counts for less at the next check. - -3: GHSA-29w2-fq35-v728 (2026-07-23, high), a security policy bypass through a startup initialisation failure in the AWS API MCP server, the server that runs arbitrary AWS CLI commands, and GHSA-2cpp-j2fc-qhp7 (2026-03-16, moderate), a file access restriction bypass in the same server. Both fixed and published, so the deduction is reduced (https://github.com/awslabs/mcp/security). - -2: five more advisories in other servers between 2026-07-14 and 2026-08-05, SSRF in HealthLake (high), improper pathname limitation in Transform (high), path traversal in the HealthOmics linters, credential disclosure via prompt injection in Amazon MQ and incorrect authorisation in the DocumentDB aggregation tool. All fixed and published (https://github.com/awslabs/mcp/security). ## 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. - Rate-limit numbers and 429 behaviour for the Knowledge server; none are published. - Whether the AWS Health Dashboard lists the Knowledge server; the page is script-rendered. - The managed AWS MCP Server (preview) docs page returned a redirect loop, so its status and controls are unchecked. - Reply times on issues; the list showed July reports still marked needs-triage. ## Weaknesses - No published rate-limit numbers, retry guidance or status component for the Knowledge server - `READ_OPERATIONS_ONLY` and `REQUIRE_MUTATION_CONSENT` default to false on the AWS API server - Seven security advisories in 2026, including a policy bypass in the AWS API server - 200 open issues, with July crash reports still awaiting triage - Per-server changelogs lag the releases, and annotations are missing on servers such as the documentation server ## 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. - Use `https://knowledge-mcp.global.api.aws` for docs and regional availability. It's rate-limited with no published numbers, so cache results - Call `get_regional_availability` before proposing a service or CloudFormation resource in a region - Run the AWS API server with `READ_OPERATIONS_ONLY=true` for inspection, and `REQUIRE_MUTATION_CONSENT=true` when writes are allowed - Use a scoped IAM role, not an admin profile. IAM is the boundary the servers rely on - Pin package versions. Releases land several times a month and can rename or remove tools ## What the review panel asked for - changelog entry per server release - removal dates on deprecations - read-only by default - injection notes everywhere ## 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.