The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to process and refactor code based on the content of comments found in the source repository. This creates a surface for indirect prompt injection, where an adversary could place malicious instructions within a code comment to influence the agent's behavior during the refactoring process.
Ingestion points: The assets/lint-comments.mjs script reads all tracked and untracked source files in the repository. The assets/fix-comment-lint/SKILL.md instructions guide the agent to interpret the purpose of each flagged comment.
Boundary markers: No explicit boundary markers or "ignore instructions" warnings are applied to the comment content before the agent processes it.
Capability inventory: The agent has the capability to write files, execute git commands, and run scripts through the project's package manager (e.g., npm, bun).
Sanitization: The skill does not perform sanitization or filtering on the comment text interpreted by the agent.
[DYNAMIC_EXECUTION]: The skill installs a custom script (assets/lint-comments.mjs) into the repository's scripts/ directory and executes it as part of the check command. The script uses node:child_process to run git ls-files for identifying source files.
[EXTERNAL_DOWNLOADS]: The skill installs several third-party development packages through the project's authoritative package manager. These include oxlint, oxfmt, jscpd, knip, and oxc-parser. These are well-known, reputable tools in the JavaScript and TypeScript development ecosystem.
[COMMAND_EXECUTION]: The skill performs various shell operations including git worktree management, git cherry-pick, and package manager invocations. It also uses the nice -n 15 command to manage the priority of long-running linting and installation tasks.

setup-ts-gate