# Fix list: letme From Anchor Terminal's listing at https://www.anchorterminal.com/tools/letme, the October 2026 research run, assessed 2 October 2026. Grade F, 36.7 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 letme: 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, 10 out of 100, up to 18 more on the total Why it scored 10: Read with the hosted checklist, and strictly, because letme is ours. No status page for letme.dev (letme.dev/status and anchorterminal.com/status answer 404, and neither /live/ nor /api/v1/workers.json lists letme), so 0 for a status page. No incident history to read, so 5 by the rule. No rate limit for picking is published, and the 5, 20 and 100 requests a second on /letme/ belong to a letme key that isn't issued, so read strictly 0. The Agent Skill at letme.dev/skills/letme/SKILL.md, linked from llms.txt, documents a 503 with Retry-After while letme is starting and says to wait and retry once, and nothing documents a 429 or how picking is limited (5 of 15, read strictly). Free with no paid tier, so no SLA (0). Every answer carries "live": false and calling.open false, so read strictly the surface isn't generally available (0). 10 in all. 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, 23 out of 100, up to 13.5 more on the total Why it scored 23: Read strictly. Credential model 0 of 30. Picking needs no credential, so there's nothing to grade, and the self-issued Ed25519 letme key on /letme/ isn't issued and earns nothing. Least privilege 12 of 20. Every live address is a read and the OpenAPI lists no write, so no destructive action exists, but there are no modes or scopes to choose and the MCP tool list is unchecked. Prompt injection 5 of 15. Answers carry auth notes, agent notes and shell or config snippets written from vendors' docs, among them a Tavily endpoint with the key in the query string (https://mcp.tavily.com/mcp/?tavilyApiKey= in the web.search answer), and nothing tells an agent to treat them as data. Operator visibility 0 of 15. No keys, no usage view and no per-call log. Security programme 6 of 20. security.txt and a disclosure policy (acknowledgement within two working days, a fix note published) on anchorterminal.com, none on letme.dev (404), no bounty, no SOC 2 or ISO 27001 and no advisories. The 2 October changelog reports an independent review of letme's POST with five medium and eleven low findings fixed, which is our own claim and earns nothing. 23 in all. 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. Agent ergonomics, 54 out of 100, up to 7.5 more on the total Why it scored 54: Context cost 12 of 25. An answer can be cut to the pick with n=0 and asked for as Markdown, but there's no field selection, and every answer repeats the directory, rule and filters blocks beside each listing's connect snippets, notes and prices. The two MCP tools would be few, but their definitions couldn't be read, so this is scored as an API. Filtering 14 of 20, read strictly. Ten filters (where, kind, auth, keyless, x402, pricing, max, minGrade, not, n), up to ten alternatives and no paging. Errors 10 of 20, read strictly. The OpenAPI and the Skill say a 400 names the bad parameter and a 404 carries suggestions and a search link, and the status codes we saw matched (a bad minGrade gave 400, /localghost gave 404), but our fetcher drops 4xx bodies, so we couldn't read one. Safe retries 10 of 20. Every address is a read with no side effect, the Skill says to wait for Retry-After and retry once on a 503, and llms.txt documents not= to fall back after a failed call. MCP readOnlyHint annotations unchecked. Defaults 8 of 15. No parameter is required on GET and n defaults to 3, but there's no SDK in any language. 54 in all. 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. ## 4. Payments & pricing, 50 out of 100, up to 6.3 more on the total Why it scored 50: Scored on the live surface only. No machine payment protocol on letme's own endpoints, and the x402 top-ups specified for when calling opens earn nothing (0 of 40). Pricing 10 of 20, read strictly. Picking is free and that's stated on letme.dev, in llms.txt and in the Skill, but letme's own price schedule for calling isn't live, so only the zero price counts. Free use with no card 20, since picking needs no card, account or key. Autonomous onboarding 20, since there's no signup and every call we made was keyless. 50 in all. 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. ## 5. Maintenance & community, 35 out of 100, up to 5.7 more on the total Why it scored 35: Read as a closed service. Last published change on 2 October 2026, the changelog entry that made letme's price step compare like with like and moved 76 of 283 picks (30). Dated letme entries on 1 and 2 October 2026 only, and the 25 September entry concerned its predecessor, Anchor Proxy, so read strictly 0 of 20. Responsiveness 5 of 15. A public changelog, contact addresses and a contact endpoint, but no public tracker, no reply seen, and the GitHub link on /letme/ and /about/ answers 404. Not in the official MCP registry (a search for letme returned letmepost and letmediff) and no SDKs (0). No public source, package or CI to judge (0). 35 in all. 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. ## 6. Transparency & trust, 40 out of 100, up to 5.3 more on the total Made of editorial 15, provenance 65. Why it scored 40: Editorial half only. Closed source, and github.com/anchorterminal answers 404 although /letme/ and /about/ link to it. The terms are three sentences on /about/#terms (data under CC BY 4.0, assessments as opinions, API use subject to the docs' rate limits), and the OpenAPI points to /letme/ as its terms of service. Neither says what letme.dev guarantees or logs for a pick, so read strictly 5 of 30. Data handling 10 of 30. /about/ says the website sets no cookies and keeps aggregate server logs rotated after 30 days, and says in the future tense that letme will log method, tool, timing, status class, size and the calling key. Nothing says what letme.dev logs for a pick today or whether job text sent in ?q= is kept. The 90-day deprecation notice on /docs/ covers /api/v1/, not letme.dev (0 of 20, read strictly). No hosting provider, subprocessors or data location named (0). 15 in all. 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: letme.dev, registered 2026-09-30 (under a year) (0 of 15) - Status page: not found (0 of 10) - security.txt: not found (0 of 10) ## 7. Schema & documentation, 68 out of 100, up to 5.2 more on the total Why it scored 68: OpenAPI 3.1 at letme.dev/openapi.json covers GET and POST /, GET /{target}, /llms.txt, /words.json and POST /mcp, with typed parameters and Error and NotFound schemas. The MCP tools pick and about couldn't be listed (GET /mcp answers 405 and our fetcher can't POST), so read strictly 20 of 25. llms.txt, the Agent Skill and Markdown answers with a .md suffix or Accept: text/markdown (10). The Skill's description says when to use letme ("Use when an agent needs a tool for a job") and four Don't lines say when not to, and each OpenAPI operation says what it resolves, with the MCP tool descriptions unseen (14 of 20). minGrade, keyless and x402 are enums and n is bounded 0 to 10, but where, kind, auth and pricing are comma-separated strings with the allowed values only in prose, and the POST body types the same fields as arrays (9 of 15). Every parameter has an example, 400 and 404 have schemas that name the bad parameter or suggest alternatives, and the Skill documents the 503 with Retry-After, but there are no response examples and no 429 (10 of 15, read strictly). The changelog has dated letme entries on 1 and 2 October 2026, but the OpenAPI version is a build stamp (2026.10.02-1614) and letme.dev has no version in the path and no compatibility promise (5 of 15, read strictly). 68 in all. 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. ## Deductions Each comes off the total. A fixed and documented problem counts for less at the next check. - 2026-10-02, -2. For speech.stt the answer's why says "price couldn't be compared (no two listings at grade BB publish a price for the job in the same unit)" while the same answer gives all four BB listings a lead price per audio-minute ($0.00167, $0.0024, $0.002 and $0.003). It then picks by overall score (77 against 73.6). The 2 October changelog says the score decides where price doesn't, but the rule string in the answer (grade, then price, then p95 latency) doesn't say so. Azure is also the cheapest of the four, so the pick wouldn't change. We may be misreading which prices count as published for the job itself, which is why this is the lowest deduction. Reproduced on directory build 20261002T161644Z. Source https://letme.dev/speech.stt.json?n=10&minGrade=BB ## 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. - unchecked: the MCP endpoint's tools/list (pick and about), their input schemas, descriptions and readOnlyHint annotations. GET /mcp answers 405 and our fetcher can't POST, so Schema and Ergonomics rest on the OpenAPI - unchecked: the bodies of 400 and 404 answers. Our fetcher reports the status and drops the body - unchecked: response headers (Cache-Control, rate-limit headers, CORS) - unchecked: whether letme rereads the directory every 15 minutes, as the listing says. Answers read on 2 October 2026 showed a build at 16:16 UTC fetched at 16:31 UTC - No rate limit for picking was found, and no statement of what letme.dev logs for a pick today - The answers don't say which prices count as published for the job itself, so the speech.stt deduction may rest on a misreading of the price step - The anchorterminal.com security.txt Expires value was the fetch time plus one year when one grader read it, which suggests it's regenerated on each request. A rolling expiry doesn't show the contact is maintained - Once these scores publish, the 'listed and not graded' statements on /letme/, in the changelog, in the listing's disclosure and notable fields, and the picker's exclusion of ungraded listings will need rewording ## Weaknesses - Calling through letme, the key and x402 top-ups are specified and not open (calling.open is false in every answer), so an agent still needs each tool's own key or x402 route - No status page, no published rate limit for picking, no 429 guidance and no SLA, and the limits on /letme/ belong to a key that isn't issued - Closed source, no SDK and no MCP registry entry, and the GitHub link on /letme/ and /about/ answers 404 - On 2 October 2026 the speech.stt answer said price couldn't be compared while listing four same-grade prices per audio-minute - Run by the company that grades the tools it picks from, with no stated logging for picks, no subprocessors and no deprecation policy for letme.dev ## 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. - Call the pick direct with its endpoint and connect snippet. letme.dev doesn't call tools yet, and calling.open says when that changes - Add not= to skip a listing after a failed call, and n=0 to drop the alternatives and shrink the answer - Check understood.confidence on a job in words, and ask again with a capability id (letme.dev/web.search) when it's low - Don't ask for ungraded listings by slug. letme.dev/letme and letme.dev/localghost answer 404 - Send keys in a header, not in the URL, even where an answer's endpoint shows ?tavilyApiKey= ## 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.