| 1 | # codacy-audit how-tos catalog |
| 2 | |
| 3 | This catalog is **live**. When an assistant performs concrete analysis on a Codacy question that requires more than one wrapper call OR more than one jq pipeline OR cross-referencing another skill, AND the answer is not already documented here or in the per-domain SKILL.md, the assistant MUST author a new entry under `how-tos/<slug>.md` AND add it here BEFORE marking the task complete. |
| 4 | |
| 5 | Skipping this rule means the next assistant repeats the analysis from scratch -- that is an explicit framework violation. |
| 6 | |
| 7 | ## Entries |
| 8 | |
| 9 | - [fetch-large-pr-issue-list](fetch-large-pr-issue-list.md) -- fix or verify the `jq: Argument list too long` failure mode by passing large Codacy issue arrays through a temporary file and `jq --slurpfile`. |
| 10 | - [handle-malformed-local-json](handle-malformed-local-json.md) -- handle `analyze-local.sh` dumps that have a `.json` suffix but contain Codacy tool-runner logs instead of parseable JSON. |
| 11 | - [reproduce-pr-22423-markdownlint](reproduce-pr-22423-markdownlint.md) -- reproduce the 864 markdownlint findings PR #22423 saw on its first CI run, locally via `analyze-local.sh --tool markdownlint`. |
| 12 | - [triage-action-required-without-token](triage-action-required-without-token.md) -- fetch public Codacy v3 PR issue details when GitHub check-run annotations are empty and no `CODACY_TOKEN` is available. |