The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and summarize external bug bounty writeups from a large index of third-party links. This creates a surface for indirect prompt injection where instructions embedded in processed articles could attempt to influence the agent's behavior.
Ingestion points: The skill reads README.md and fetches content from external URLs referenced within it.
Boundary markers: The skill contains a 'Safety Boundary' section instructing the agent to confirm authorization and provide non-destructive advice, which acts as a partial behavioral constraint.
Capability inventory: The agent generates testing hypotheses and can fetch URL content to inform its output.
Sanitization: No explicit sanitization or filtering of external content is specified before processing.
[EXTERNAL_DOWNLOADS]: The skill instructions encourage the agent to fetch and analyze data from hundreds of external security blogs, research sites, and GitHub repositories listed in the bundled index.
[OBFUSCATION]: A homoglyph was identified in README.md within a URL path. Specifically, the character 'с' in the word 'cookie' in one link is a Cyrillic character (U+0441) rather than a Latin 'c'. This appears to be a clerical artifact in the link text/path rather than a malicious redirection attempt, as the domain remains the well-known 'medium.com'.

bugbounty-ideas