The Agent Skills Directory

[SAFE]: The skill operates as intended for project management tasks, defining clear workflows for research and estimation. It leverages specific MCP bridge tools and common file access tools for architectural analysis and documentation lookups.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core function of reading and processing untrusted codebase data and external documentation.
Ingestion points: Analyzes local repository content via Read, Grep, and Glob tools and retrieves external library documentation using the mcp_context7 tool.
Boundary markers: No specific delimiters or instructions are provided to the agent to help it distinguish between data from the codebase/documentation and its own system instructions.
Capability inventory: The agent is authorized to use the Bash tool to execute verification commands and utilizes multiple MCP bridges (gemini-bridge, open-bridge, codex-bridge) for generating plans.
Sanitization: There is no indication that data retrieved from external sources or the codebase is sanitized or validated before being included in planning phases or execution blocks.

project-planning