About io.github.cyanheads/evals-mcp-server
io.github.cyanheads/evals-mcp-server is an MCP server in the Security category: author verifiable eval records through a draft→review→revise→submit loop with enforced graders.
Install
claude mcp add io-github-cyanheads-evals-mcp-server -- npx -y @cyanheads/evals-mcp-servernpx -y @cyanheads/evals-mcp-serverThis server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.
Tools (9)
evals_create_draftCreate a draft eval record carrying its own grader; returns the parsed record, a review protocol, and a verification subagent prompt.
Parameters not published.
evals_describe_schemaReturn the required and optional fields plus grader options for a task type. Call before drafting.
Parameters not published.
evals_discard_draftDelete a draft record by id. Draft-only.
Parameters not published.
evals_export_recordsCompile submitted records to JSONL, CSV, Inspect AI, or lm-evaluation-harness and write the artifact under `exports/`.
Parameters not published.
evals_get_recordRead a draft or submitted record by id; the id is stable across submit.
Parameters not published.
evals_list_recordsBrowse and filter records by status, domain, task type, or tag. Returns a compact summary per record.
Parameters not published.
evals_revise_draftApply a surgical `set` / `append` / `unset` patch to a draft by dotted path; re-runs the self-consistency check.
Parameters not published.
evals_run_checkRun a grader spec against candidate answers and get PASS/REJECT per candidate, decoupled from any saved record.
Parameters not published.
evals_submit_draftFinalize a draft through the committability gate, then freeze it.
Parameters not published.
Ask AI
Ask AI about io.github.cyanheads/evals-mcp-server
Powered by Claude · Grounded in docs
Security checks
6 checks have not been run yet.