makinari-rol-judge-doc

To automate markdown lint checks for doc approval flows, you submit a doc along with a lint/markdown tool-call as evidence. The flow then validates this evidence and provides a deterministic verdict for approval.

Evidence-based doc judging is a process that enforces a lint/markdown validation step in approval flows. It works by requiring a lint tool-call as evidence and verifying acceptance tokens within that evidence haystack before approving documents.

You integrate automated quality gates into CI-like pipelines by providing a doc submission with a lint/markdown tool-call in the evidence. This setup ensures consistent approvals by validating markdown content and evidence deterministically.

You need a lint/markdown tool-call included in your evidence to verify acceptance tokens. The judging mechanism checks for these acceptance tokens within the evidence haystack, which includes the lint output, commit summary, and claim.

Your markdown doc flow fails to get an approved verdict if the evidence lacks a required lint/markdown tool-call or if the acceptance tokens are not present within the evidence haystack. This prevents unchecked documents from proceeding.

Yes, you can use this automated doc judging for per-flow doc tasks. It applies deterministic quality gates to validate markdown content and evidence, ensuring consistent approvals across individual document submissions.