The Agent Skills Directory

[SAFE]: The security analysis did not find any malicious patterns, prompt injections, or unauthorized data exfiltration attempts. The skill follows best practices for tool execution and data handling.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes meeting transcripts and chat threads. However, it implements robust safeguards, including explicit instructions to the agent in SKILL.md and frameworks/transcripts.md to treat these sources as untrusted data rather than instructions.
[COMMAND_EXECUTION]: The skill utilizes local shell and PowerShell scripts (aa.sh, init.sh, lab-probe.sh, etc.) to automate project management tasks. These operations are limited to the project directory, do not require elevated privileges, and the skill instructions mandate user approval for every command executed.
[DATA_EXFILTRATION]: No exfiltration patterns were detected. The skill includes a proactive defensive tool, transcript-normalize, which redacts credentials (API keys, tokens), PII (emails, phone numbers), and financial data (IBANs, credit cards) from input files using a comprehensive regex suite before they are stored or processed.

agentic-agile