The Agent Skills Directory

[SAFE]: The skill focuses on legitimate development tasks including git conflict resolution, build error fixing, and changeset management. It uses standard development tools (git, pnpm) and includes explicit instructions to avoid risky behavior, such as preventing the re-installation of dependencies during CI fixes to maintain lockfile integrity.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from git diffs, changeset files, source code, and CI logs, which could theoretically contain adversarial content. However, the risk is inherent to the skill's primary purpose of repository management, and its actions are constrained to standard build/test/git workflows.
Ingestion points: Git diff outputs, changeset markdown files, project source code, and ciLogs argument referenced in clean-changesets.md, fix-ci.md, and resolve-conflicts.md.
Boundary markers: None present; the skill relies on the agent's internal reasoning.
Capability inventory: git add, git rm, git checkout, pnpm build, and pnpm test executions.
Sanitization: No explicit sanitization or filtering of external logs or diff content is specified.

merge