The Agent Skills Directory

[SAFE]: The skill serves as a behavioral guideline for communication and does not contain executable code, scripts, or remote dependencies. All instructions align with the stated purpose of improving technical communication.
[INDIRECT_PROMPT_INJECTION]: The skill instructions involve reading and processing external data from files, configurations, and command outputs to verify tasks.
Ingestion points: SKILL.md specifies gathering facts from environment-discoverable files and command outputs.
Boundary markers: The skill does not define specific delimiters for untrusted data, though it emphasizes structured reporting.
Capability inventory: The skill references the use of CLI tools (e.g., pytest, uv) and file system operations as part of its verification workflow.
Sanitization: No explicit sanitization or escaping mechanisms for untrusted data are defined in the instructions.
[COMMAND_EXECUTION]: The skill provides examples of using development tools (e.g., uv, pytest) to verify work. These are presented as best-practice examples for technical task reporting rather than automated or hidden command execution.
[CREDENTIALS_SAFE]: No hardcoded credentials, API keys, or instructions to access sensitive credential files (like .aws/credentials or .ssh/id_rsa) were detected.
[DATA_EXFILTRATION]: No network communication patterns, non-whitelisted domains, or data exfiltration vectors were identified.

expression-skill