The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a Python script (scripts/ship_gate_scanner.py) which the agent is instructed to run to perform project audits. This script is well-structured, uses standard Python libraries for file system traversal and regex-based scanning, and does not execute external shell commands or untrusted code.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it reads and displays content from the files it audits. This could allow malicious content within a scanned file to attempt to influence the agent's behavior. However, the risk is mitigated by the skill's specific purpose, the truncation of displayed snippets (120 characters), and the agent's role as the final arbiter. Mandatory Evidence:
Ingestion points: scripts/ship_gate_scanner.py reads local files via the grep_files and walk_files functions.
Boundary markers: Absent; results are returned as plain text findings.
Capability inventory: File system read access and console output to the agent.
Sanitization: Snippets are limited to 120 characters in the scanner script.
[SAFE]: The skill lacks network capabilities and does not attempt to exfiltrate any data found during the audit process. No hardcoded credentials or obfuscation techniques were identified.

ship-gate