The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes local command-line tools to manage the discovery workflow and state.
Command List: It uses mkdir for directory setup, ao for searching prior session history, and bd for task/epic management.
Vendor Scripts: The skill executes local scripts scripts/checkpoint-commit.sh and scripts/log-telemetry.sh, which are part of the author's environment for persistence and monitoring.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it processes user input and generated content.
Ingestion points: Untrusted data enters via the <goal> parameter in SKILL.md and through pre-mortem reports read from the .agents/council/ directory.
Boundary markers: No explicit delimiters or 'ignore' instructions are used when interpolating these variables into shell commands or subsequent skill invocations.
Capability inventory: The skill can execute local shell commands and call other functional skills within the agent environment.
Sanitization: There is no evidence of input escaping or validation for the data ingested during the workflow orchestration.

discovery