The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from GOALS.md or legacy YAML files to perform fitness measurements.
Ingestion points: Reads GOALS.md or alternative YAML project goal files from the project directory.
Boundary markers: Explicitly instructs the agent to avoid "advice creep" (turning reports into work selection) and provides a specific format for results, requiring the agent to stop after returning the aggregate measurement.
Capability inventory: The skill executes ao goals CLI commands and writes data to the filesystem at .agents/ao/goals/baselines/ or a caller-specified path via the render --out flag.
Sanitization: No explicit sanitization or content validation of the input files is described before processing.
[COMMAND_EXECUTION]: The skill relies on the ao goals CLI family (measure, validate, drift, history, export, meta, scenarios, render) to perform its core functions. These commands are executed via the shell to generate JSON or text outputs.

fitness