The Agent Skills Directory

Command Execution for Repository Analysis: The skill executes standard system commands such as find, awk, wc, and sed to analyze the project's file structure, line counts, and directory depth.
These commands are used solely for gathering metadata about the codebase to determine where documentation should be generated.
Indirect Prompt Injection Surface: The skill reads existing documentation files (like AGENTS.md and CLAUDE.md) and source code comments to inform its generation process.
Ingestion points: Content is retrieved via Read operations and find commands across the local filesystem (SKILL.md).
Boundary markers: The instructions do not currently specify explicit delimiters to separate processed data from agent instructions.
Capability inventory: The skill has the ability to execute bash commands, spawn subagents via Task calls, and write files to the disk (SKILL.md).
Sanitization: There is no explicit sanitization of the content read from the files before it is processed by the agent.
Note: While this presents a surface for potential instruction influence from the codebase being analyzed, it is a common characteristic of automated documentation tools.
Dynamic Agent Spawning: To handle large projects efficiently, the skill dynamically spawns subagents using Task calls to perform parallel analysis tasks.
This architecture is a standard practice for managing complexity in agentic workflows and allows the tool to scale with the project size.

index-knowledge