The Agent Skills Directory

[SAFE]: The skill serves as an instructional guide for developers using Pathfinder MCP tools. It contains standard engineering workflows for debugging, implementing features, and refactoring without any malicious logic, exfiltration, or unauthorized access patterns.
[PROMPT_INJECTION]: The skill facilitates the reading and processing of external codebases via tools like read_source_file and get_repo_map, which introduces a surface for indirect prompt injection if the files being analyzed contain malicious instructions designed to influence the agent.
Ingestion points: Code content is ingested through read_symbol_scope, read_source_file, read_with_deep_context, search_codebase, and get_repo_map as described in SKILL.md.
Boundary markers: No explicit boundary markers or 'ignore' instructions for code content are specified in the provided workflows.
Capability inventory: The agent has the capability to execute shell commands via Bash (e.g., for tests) and modify or delete files using replace_body, create_file, and delete_file tools.
Sanitization: There is no mention of sanitization or filtering for the code being read before it enters the agent's context.

pathfinder-workflow