The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs shell command execution to manage isolated audit environments and verify code quality.
Evidence: The workflow in SKILL.md utilizes git worktree add, git checkout, and git merge to create a dedicated workspace for auditing, preventing direct modification of the main branch.
Evidence: SKILL.md instructs the agent to execute project-specific verification commands, such as npm test, npm run build, and npm run lint, which are defined by the user or project configuration.
Evidence: A shell loop is used in SKILL.md to run tests 100 times to verify the stability of property-based tests.
[PROMPT_INJECTION]: The skill's architecture is susceptible to indirect prompt injection due to its processing of external code files.
Ingestion points: The skill accepts a file or directory path and reads TypeScript source files as input for analysis, as described in README.md.
Boundary markers: The skill implements an 'Interactive Change Approval Pattern' that presents a summary table and detailed code diffs to the user before applying changes, acting as a manual check.
Capability inventory: The skill possesses file system write access (creating tracking files in .agents/audit/), git command execution, and the ability to run arbitrary project scripts.
Sanitization: No specific sanitization of the audited file content is mentioned; the safety of the operation relies on the human-in-the-loop approval process.

accelint-ts-audit-all