The Agent Skills Directory

[SAFE]: The skill is composed of instructional markdown documents and configuration files. A thorough review of all 34 files confirms the absence of prompt injection patterns, hardcoded credentials, malicious persistence mechanisms, or obfuscated payloads.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided code for review and refactoring workflows, which presents a surface for potential indirect prompt injection attacks.
Ingestion points: User-provided code snippets or file contents processed by the 'Reviewer' (workflows/reviewer.md) and 'Refactor' (workflows/refactor.md) workflows.
Boundary markers: None explicitly defined in the instructions to separate user data from agent instructions.
Capability inventory: The skill allows the agent to write markdown files to the local file system (Architecture Decision Records) and conditionally chain to a 'humanizer' skill.
Sanitization: No specific sanitization or filtering logic is specified for the processed code.
[COMMAND_EXECUTION]: The 'Auto-ADR' feature instructs the agent to create and save Architecture Decision Records as markdown files in the docs/adr/ directory. This file system write operation is a legitimate functional requirement of the engineering coach workflow and is documented as intended behavior in SKILL.md and checklists/adr.md.

compass