# Fix list: BlockRun.AI From Anchor Terminal's listing at https://www.anchorterminal.com/tools/blockrun-ai, the October 2026 research run, assessed 1 October 2026. Grade BB, 72.5 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 BlockRun.AI: 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, 55 out of 100, up to 9 more on the total Why it scored 55: Graded on the hosted gateway, read as a router, since every MCP and SDK call lands there. blockrun.ai/status is BlockRun's own page with live checks of the Base and Solana gateways and 13 upstream providers, but it keeps no history and says 'None recorded on this page yet' (10 of 20, and 5 for no readable incident record). Rate limits published with numbers, 30 requests a minute and 300 an hour per IP on free models and 60 to 120 an hour on metadata endpoints, with no platform cap on paid inference (15). 429s carry Retry-After, X-RateLimit-Source and a RATE_LIMITED body, with backoff and failover guidance, settlement only after a successful upstream response and a PAYMENT_REPLAY check on reused nonces (15). The terms say 'We do not commit to any particular uptime' (0). x402 v2 runs on Base mainnet in production (10). 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, 70 out of 100, up to 5.3 more on the total Why it scored 70: Model reading. The credential is a wallet private key, auto-created at `~/.blockrun/.session` or passed in an environment variable, and it controls the whole balance. The MCP wallet tool can delegate a capped budget to a child agent, paid calls prompt for approval above a threshold, and an API-key route exists. 20 of 30, a judgement call. The privacy policy (updated 2026-04-28) says 'We do not train models on your data' and that BlockRun uses providers' paid API tiers (15 of 15). Prompts and completions 'are not stored beyond the duration of the request' and request metadata is kept 30 days, while upstream provider retention is left to each provider's terms (15 of 20). Every payment settles on-chain and the wallet tool reports spending (10 of 15). security.txt valid per the listing's provenance check, and the MCP repository's SECURITY.md routes reports through GitHub private reporting with a 72-hour acknowledgement. No bounty or certification found (10 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. ## 3. Agent ergonomics, 70 out of 100, up to 4.9 more on the total Why it scored 70: Model reading of the checklist (tool use, structured output, caching, context, batch, SDKs, errors). Tool calls pass through the OpenAI-compatible endpoint, not documented by BlockRun beyond that (10 of 20). `response_format: json_schema` structured output supported since 2026-06-28 per the changelog (15). A response cache in the SDK so a retried call isn't paid twice, and cache tokens reported in usage since 2026-06-22, no prompt-caching discount of its own (5 of 15). Routes to models with 1M-token context (15). No batch API (0). Official SDKs in Python, TypeScript and Go (10). Errors documented with machine-readable codes, whether each status is charged, and what to do next (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. ## 4. Schema & documentation, 75 out of 100, up to 4.1 more on the total Why it scored 75: The vendor's llms.txt names an OpenAPI 3.1 document at blockrun.ai/api/openapi, and the MCP repository's changelog cites it as the live route list, but robots.txt blocked our reader so we couldn't open it (10 of 25, a judgement call). llms.txt served at blockrun.ai/llms.txt (10). The 19 MCP tool descriptions in the source state purpose and order of use, such as 'discover markets ... with blockrun_markets first' and 'Run action:"setup" FIRST' (15 of 20). Inputs are zod schemas with enums and numeric bounds, but seven fields across six tools take a free-form `z.any()` JSON body (10 of 15). The docs have an error reference with codes per status and recovery steps, and the README has examples for each client (15). A dated gateway changelog, last entry 2026-09-29, and a CHANGELOG.md in the MCP repository (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. ## 5. Transparency & trust, 66 out of 100, up to 3 more on the total Made of editorial 50, provenance 82. Why it scored 66: SDKs, the MCP server and ClawRouter are MIT, the gateway is closed, and the terms name BlockRun, Inc., a Delaware corporation in San Francisco (15). The privacy policy states no training, no storage of prompts past the request and 30 days for request metadata. It names no subprocessors, and its footer says 'BlockRun Labs, Inc.' while the terms say 'BlockRun, Inc.' (20 of 30). No deprecation policy, and the terms say a provider 'may change, deprecate, rate limit, or withdraw a model at any time'. Removals are dated in the changelog (5 of 20). Upstream providers are named and Google Analytics is disclosed for page views. Data locations aren't stated (10 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: blockrun.ai, registered 2025-10-06 (under a year) (0 of 15) ## 6. Maintenance & community, 85 out of 100, up to 1.3 more on the total Why it scored 85: Model reading. Last gateway changelog entry 2026-09-29, and @blockrun/mcp v0.53.1 tagged the same day (30). Removals are dated in the changelog, GPT-5.3 on 2026-08-29 and four free NVIDIA models on 2026-08-30, but on the day rather than with notice (4 of 12). The retired GPT-5.3 redirects to GPT-5.2 with no code change, so calls didn't break (6 of 8). A public, dated changelog with entries most weeks (15). Issue replies not sampled, since GitHub's issue pages are closed to our reader (5 of 10). Official SDKs in Python, TypeScript and Go (15). CI on Node 20.19 and 22 with typecheck and tests on every pull request, and Renovate for dependencies (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. ## 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 OpenAPI document at blockrun.ai/api/openapi, which robots.txt keeps our reader out of. Named in llms.txt and the vendor's repository - unchecked: issue and pull-request response times on GitHub, whose issue pages are closed to our reader - The terms name BlockRun, Inc. and the privacy policy footer names BlockRun Labs, Inc. Which entity processes data isn't stated - No subprocessor list or data locations, and how long upstream providers keep prompts is left to their own terms - The source registers 19 MCP tools, as the listing first said. Yesterday's patch to 18 was wrong and is reversed - Graded with the model reading for consistency with the batch, although the listing's kind is mcp ## Weaknesses - The status page shows live checks only and keeps no incident history - The terms say 'We do not commit to any particular uptime', and there's no SLA - Models are removed on the day, with no notice period, and the terms let providers withdraw a model at any time - No subprocessor list or data locations, and the privacy footer names a different entity from the terms - The wallet key controls the whole balance and sits in a local file or environment variable ## 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. - Fund a dedicated wallet with a small USDC balance. The balance is the spending cap - Use `blockrun_wallet` delegate with `agent_limit` before handing a sub-agent the server - On a 429, wait `Retry-After` and read `X-RateLimit-Source` to fail over to the same tier on another provider - Call `blockrun_models` before hard-coding a model name, since models are removed without notice - Use `mode:"free"` for drafts and classification, paid models for generation ## What the review panel asked for - a minimum notice period before removals - dated model retirement notices - Explain 402 amount setting - State the card route's minimum ## 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.