The Agent Skills Directory

[SAFE]: The skill operates as a code analysis tool using standard commands like Grep, Glob, and Read to map project structures and extract business logic. All analyzed logic is confined to documentation generation and architecture mapping.
[EXTERNAL_DOWNLOADS]: The skill references documentation at https://jeffallan.github.io/claude-skills/. While automated scanners flagged these as malicious, manual inspection shows they belong to the author's GitHub Pages site for documentation hosting. There is no evidence of automated downloads or remote code execution from these URLs; they are provided as informational links for the user.
[COMMAND_EXECUTION]: The skill uses Bash, Grep, and Glob tools. These are used strictly for local file discovery (e.g., finding .py or .ts files and searching for strings like TODO or app.route). No dangerous command patterns or privilege escalation attempts (like sudo) were identified.
[INDIRECT_PROMPT_INJECTION]: The skill processes existing codebases. It contains clear instructions for the agent to distinguish between observed facts and inferences and provides structured templates (EARS format) to reduce the risk of following instructions embedded within analyzed code files.

spec-miner