Evidence locator integrity

Reusable AI agent skills, project templates, and guardrails for safer software maintenance and delivery.

Install

npx -y skills add jukrap/ai-agent-playbook --skill evidence-locator-integrity

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

Copied from the file, not written here

Use when claims need reopenable locators, scan ranges, source boundaries, or evidence caveats.

1.4 KB, as published. Nobody here has run it

Use this as the primary AI harness skill for checking whether evidence can be reopened before a claim is trusted, cited, or promoted.

Classify the claim as code behavior, structure, absence, risk, runtime report, source registry entry, external source, command output, data result, or manual observation.
Attach a locator that can be reopened from the target context: path range, symbol, runtime artifact, source registry item, command output, URL, issue, database/query, or manual observation record.
For absence, coverage, or "no usages found" claims, record the scan range, tools used, skipped paths, and freshness before treating the claim as evidence.
Reject or soften evidence that contains local absolute paths, credential-looking values, private endpoints, uncapped excerpts, stale generated summaries, or missing source boundaries.
Keep generated evidence under runtime until a reviewed promotion step explicitly moves stable facts into memory, canon, docs, or handoffs.

Read references/locator-contract.md for reusable locator shapes, required fields, and source-boundary rules.

Read references/claim-scan-range-rules.md for claim types, scan range requirements, absence-claim rules, and unsafe evidence anti-patterns.