The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes standard local development scripts (npm run ci and npm run ci:nightly) to perform quality checks. These operations are limited to the project's own command definitions.
[PROMPT_INJECTION]: The skill processes output from various development tools (tsc, eslint, cspell, knip, etc.) to generate a report. While this constitutes a surface for indirect prompt injection, the risk is significantly mitigated by the disable-model-invocation configuration, which prevents the agent from interpreting the ingested data as new instructions.
Ingestion points: The skill captures and reads the stdout/stderr from npm run commands to parse results from files like reports/quality/QUALITY_REPORT.md.
Boundary markers: None explicitly defined in the instructional steps.
Capability inventory: File reading (Read), pattern matching (Grep), file discovery (Glob), and command execution limited to npm run (Bash).
Sanitization: The skill summarizes tool output into a markdown report; no specific sanitization or escaping of the tool output is mentioned.

quality-check