The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several local bash scripts (render.sh, compile-html.sh, compile-pdf.sh, and verify-citations.sh) to automate the processing of architectural data and the generation of final reports. These scripts employ standard developer utilities including pandoc, grep, find, sed, and awk. The execution is scoped to the report output directory and follows best practices, such as input validation for style and theme parameters.
[SAFE]: The skill implements a 'Verification Protocol' in references/verification-protocol.md that mechanically validates citations and evidence returned by sub-agents. By using low-level tools like grep and Read to confirm that cited code exists and matches the analysis, the skill effectively mitigates the risk of hallucinations or data poisoning from the codebases being analyzed. No unauthorized network access, data exfiltration patterns, or obfuscation techniques were found across the instructions, scripts, or assets.

architectural-analysis