The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines shell hooks in hooks/hooks.json that trigger the execution of the management script scripts/brainstorm-run.mjs using sh -c and Node.js during specific lifecycle events.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from multiple web search engines and external documents, creating a surface for injection.
Ingestion points: Web search results and external resources referenced in references/web-search-workers.md and references/tools.md.
Boundary markers: The skill instructs the agent to treat findings as leads and perform independent validation, but lacks technical delimiters for the ingested data.
Capability inventory: The skill has the capability to write local files for session management and execute shell commands through the hook system.
Sanitization: Instructions mandate a "stress-test" and perspective review to validate claims, which serves as a logical guardrail against untrusted input.

octocode-brainstorming