The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes source code, README files, and documentation to identify gaps and generate content. If these files contain malicious instructions (e.g., hidden in comments or prose), they could potentially influence the agent's behavior during the analysis or generation phase.
Ingestion points: Local source files, READMEs, and .env files (referenced in SKILL.md and reference/documentation-patterns.md).
Boundary markers: Absent.
Capability inventory: The agent can modify local files and execute shell commands for auditing.
Sanitization: Not explicitly mentioned for ingested content.
[DYNAMIC_EXECUTION]: The documentation provides templates for auditing scripts, such as scripts/doc-coverage.ts (using ts-morph) and scripts/type-audit.sh, which are intended to be created and executed on the local filesystem to measure documentation health. Evidence: reference/coverage-audit-tools.md.
[EXTERNAL_DOWNLOADS]: The skill recommends the installation and use of various well-known NPM packages and tools for documentation generation and auditing, including typedoc, swagger-jsdoc, markdown-link-check, type-coverage, and ts-morph. Evidence: reference/api-doc-generation.md, reference/coverage-audit-tools.md.
[COMMAND_EXECUTION]: The skill's reference guides include numerous examples of shell commands for installing dependencies, running linters, and executing the provided audit scripts. The agent may be tasked with running these commands to provide the requested coverage reports. Evidence: reference/coverage-audit-tools.md.

quill