The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external content such as error messages, log files, and API response bodies, which creates a vulnerability to instructions embedded in that data.\n
Ingestion points: User-provided error messages and diagnostic outputs from network tools like curl and ping in SKILL.md and various files in the references/ directory.\n
Boundary markers: The skill does not define clear delimiters or instructions to ignore embedded commands in the processed text.\n
Capability inventory: The skill has access to networking and system tools including curl, nc, ping, nslookup, and wc.\n
Sanitization: No sanitization or validation of external diagnostic content is performed before processing.\n- [DATA_EXFILTRATION]: The skill guides the agent to use environment variables for sensitive authentication tokens in shell commands, which could expose these secrets in process lists, command history, or to documentation endpoints.\n
Evidence: Instructions to use $API_KEY with curl and echo appear in references/issue-authentication-failed.md and references/issue-invalid-json-response.md.

troubleshooting-guide