The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool for Phase 0 intent analysis to index the project and estimate scope via Grep and Glob. It also employs the Agent tool to dispatch specialized 'Explore' sub-agents in parallel waves. This usage is consistent with the skill's primary purpose of code research.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it reads content from the user's codebase and includes synthesized facts or findings in subsequent prompts for specialized agents.\n
Ingestion points: Files are read via the Read and Grep tools during parallel exploration waves defined in SKILL.md.\n
Boundary markers: The references/agent-prompt.md file includes an explicit warning instructing agents to treat previous facts as hypotheses and to verify them independently by reading the actual code.\n
Capability inventory: The orchestrator and its sub-agents have access to Bash, Agent, Read, Grep, and Glob tools.\n
Sanitization: There is no explicit sanitization or escaping of the codebase content before it is processed by the synthesis algorithm or passed to sub-agents.

deep-explore