Run every knackpack reviewer that applies to one target together, then merge what they report into one triaged report: duplicates grouped under one ID with every source kept, severities reconciled, each serious merged finding checked once by the adjudicator, including whether its fix works, and every issue sorted into fix now, decision needed or won't fix, with how each fix would be verified. Report-only: it never edits the project's files, and it saves the report to a private temporary folder and gives the path. Use only when the user asks to run several or all of knackpack's reviewers together, or for one review by every reviewer, never automatically after a change; for a single reviewer, start that reviewer instead. It starts many agents, and on a large code base it can use more than a million tokens.
91%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
Low
Low-risk findings worth noting
The skill exposes the agent to untrusted, user-generated content from public third-party sources, creating a risk of indirect prompt injection. This includes browsing arbitrary URLs, reading social media posts or forum comments, and analyzing content from unknown websites.
The skill orchestrates multi-agent reviews of code bases and pull requests, consuming reviewer reports and potentially pull request files via CLI (`gh pr view`), which can include outsider-authored text from untrusted pull requests.
bezmoradi/knackpack
9fcf73c
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.