
Check a draft's numbers against your facts.md and its links before you publish.
Public tool metadata for what this MCP can expose to an agent.
check_numbersChecks that every number in a draft is in your facts.md or carries a [src: URL].Checks that every number in a draft is in your facts.md or carries a [src: URL].
No parameter schema in public metadata yet.
check_linksChecks a draft for dead links and checkout links without utm_source.Checks a draft for dead links and checkout links without utm_source.
No parameter schema in public metadata yet.
Sponsors
The D.O.E. framework (Directive / Orchestration / Execution) for running AI coding agents — Claude Code, Cursor, Codex, Gemini CLI — reliably on real work.
The problem: an LLM is probabilistic; your business logic isn't. If the agent executes every step itself, errors compound — 90 % per step is 59 % over five steps. D.O.E. moves the deterministic work into scripts and leaves the agent one job: deciding what to run next. Then it makes the system self-annealing: every failure becomes a test, a rule, and a lint.
The plugin lives in plugins/agent-ops-kit-lite/; the MCP server in mcp/ is separate.
https://github.com/skibuff22/agent-ops-kit-lite/releases/download/mcp-v0.1.0/publish-gate.mcpb