The Agent Skills Directory

[SAFE]: The skill performs local, read-only scans of source code to identify personally identifiable information (PII) markers and telemetry libraries for release hygiene purposes.
[SAFE]: The analysis script bin/check.sh implements explicit file path filtering to avoid reading sensitive environment and credential files such as .env or .env.local.
[INDIRECT_PROMPT_INJECTION]: The skill ingests content from project source files to identify privacy signals.
Ingestion points: The skill reads files located in src/, app/, pages/, server/, api/, and lib/ directories.
Boundary markers: Absent. The skill does not use specific delimiters to isolate untrusted file content within its internal processing.
Capability inventory: Limited to read-only file access and local JSON output generation; no network operations or arbitrary file write capabilities were detected.
Sanitization: Present. The skill uses the jq utility to safely format findings into JSON, preventing potential injection attacks that could arise from malicious file content.

privacy-check