The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses node:child_process.execFileSync to invoke the curl binary as a fallback transport mechanism for network requests. The implementation utilizes an argument array rather than a shell string, ensuring safe execution of the external binary.\n- [EXTERNAL_DOWNLOADS]: Fetches data from authoritative domains such as api.quran.com, sunnah.com, openlibrary.org, api.crossref.org, and wikipedia.org to retrieve religious and bibliographic text. It also downloads datasets from a GitHub repository (fawazahmed0/hadith-api) for content-indexed verification.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes data from external APIs and HTML scraping, creating a surface for potential indirect prompt injection attacks.\n
Ingestion points: External content is fetched from academic and religious reference providers as defined in scripts/lib/providers.mjs and processed in scripts/verify.mjs.\n
Boundary markers: The skill applies explicit normalisation profiles (e.g., arabic, latin, numeric) and match verdicts to validate retrieved text against expected values or independent corpora.\n
Capability inventory: The skill is capable of performing network fetches and writing verification records and human-readable summaries to the local filesystem.\n
Sanitization: Logic in scripts/lib/normalize.mjs (the stripTags and clean functions) handles HTML entities and performs tag stripping to clean external data before processing.

verification-agent