The Agent Skills Directory

[SAFE]: No malicious instructions, obfuscated patterns, or data exfiltration vectors were identified across the 162 files analyzed.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to perform maintenance tasks such as indexing documentation and searching specifications (e.g., node scripts/generate-index.js). The use of shell commands is inherent to the skill's purpose as a requirements manager and is limited to local script execution.
[PROMPT_INJECTION]: A surface for indirect prompt injection exists because the skill reads and processes local documentation files from the references/ directory. However, this is the intended core functionality, and no markers of adversarial intent were found.
[CREDENTIALS_UNSAFE]: While the documentation files discuss sensitive environmental variables (e.g., ANTHROPIC_API_KEY, TURSO_AUTH_TOKEN), all instances are placeholders or format descriptions for developer guidance and do not contain actual secrets.

aiworkflow-requirements