The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill scans the codebase across various dimensions (File Anatomy, State, Infrastructure, Error Handling) to extract and codify style rules. Maliciously crafted content within the scanned legacy code, such as instructions hidden in comments or string literals, could potentially influence the agent's reasoning or the contents of the generated rules file.
Ingestion points: Full-scan and incremental sniffing of project files using Read, Glob, Grep, and Bash commands.
Boundary markers: Not explicitly defined for the scanning and data ingestion phase.
Capability inventory: The skill possesses Bash for command execution, Write and Edit for file modification, and Read for data ingestion.
Sanitization: The skill implements a "Grilling Protocol" that acts as a human-in-the-loop mechanism, requiring the user to resolve strong-signal conflicts and validate patterns before they are added to the rules file.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute git commands (e.g., git ls-files, git diff, git log) to analyze project scale and track changes. This is consistent with its stated purpose of codebase alignment.
[PERSISTENCE]: The skill offers an optional "Hard hook" mechanism that modifies the IDE's settings.json to include a PreToolUse hook for the rules file. This ensures the enforcement logic persists across sessions. This action is gated by the AskUserQuestion tool to ensure user consent.

inherit-legacy-style