The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes bounty titles and descriptions from the external API at bounty.drx4.xyz, which are then used to influence agent behavior.
Ingestion points: The fetchBounties and fetchBountyDetail functions in bounty-scanner.ts retrieve bounty data from an external source.
Boundary markers: No delimiters or safety instructions are used when processing these strings.
Capability inventory: According to AGENT.md, the agent uses the matching confidence scores to decide whether to autonomously claim a bounty.
Sanitization: The skill performs basic keyword matching but does not sanitize the content against prompt injection attempts aimed at influencing the autonomous claim logic.
[DATA_EXFILTRATION]: The skill performs network operations to the non-whitelisted domain bounty.drx4.xyz.
The API endpoint is hardcoded as a default but can be overridden by the BOUNTY_API_URL environment variable, which could allow redirecting traffic to an arbitrary external server.
The skill also reads local files (skills.json and all SKILL.md files in subdirectories) to identify agent capabilities, though these are not considered sensitive system files.
[COMMAND_EXECUTION]: The skill defines a CLI interface using the commander package, providing several subcommands (scan, match, claim, detail, status, my-bounties) for the agent to execute in its environment.

bounty-scanner