harnessloop-evidencelisted

# Harnessloop Evidence Manage evidence contract changes during a Harnessloop run. This skill controls whether evidence can support acceptance; it does not execute the business task or accept a round by itself. ## Input Contract Accept an explicit skill invocation such as `$harnessloop-evidence add`, `$harnessloop-evidence check`, `$harnessloop-evidence revise`, or `$harnessloop-evidence reject`, or natural language asking to update accepted evidence. Treat `harnessloop:evidence` as a natural-language alias only; `$harnessloop:evidence` is not a valid skill invocation. Useful input includes: - `action`: `add`, `check`, `revise`, `reject`, or `diff`. - `target`: `.harnessloop/state/evidence-index.md`, a goal `data-contract.md`, a round evidence directory, or a specific evidence file. - `evidence-id`: stable ID when updating an existing evidence entry. - `evidence-type`: `static`, `dynamic`, `runtime`, `source`, or `human-confirmation`. - `path-or-url`: local file path, command output path, CI/build URL, report path, dataset path, source file, or monitoring/probe reference. - `applies-to`: project setup, goal, threshold, round, scope-lock, review, or decision. - `freshness-rule`: expected timestamp, max age, refresh cadence, or immutability statement. - `validation-method`: command, test, checksum, query, review method, access check, or reproducibility condition. - `required-tool`: required when evidence collection or validation depends on a specified tool call; include too