The Agent Skills Directory

[PROMPT_INJECTION]: The skill ingests research questions from user-provided arguments and local project files to generate prompts for sub-agents, creating a surface for indirect prompt injection. * Ingestion points: Research questions are read from $ARGUMENTS and context is gathered from files like CLAUDE.md. * Boundary markers: No explicit markers are used to isolate untrusted input within sub-agent instructions. * Capability inventory: The skill can spawn sub-agents (Task tool), write to the filesystem, and execute specific shell commands. * Sanitization: Input content is processed directly without specific escaping or validation.
[COMMAND_EXECUTION]: The skill uses a command-line interface to invoke multiple model providers in 'Council Mode'. * Evidence: It executes 'uv run python -m cli_council' within a local package directory (packages/cli-council) using temporary files for data transfer.

multi-perspective