The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a local script to validate state transitions during the execution process.
Evidence: SKILL.md Step 3.1 instructs the agent to run python3 scripts/validate-transition.py <from> <to>.
The script is included in the skill package and uses basic string comparison to verify transition logic.
[INDIRECT_PROMPT_INJECTION]: The skill interacts with external task state data, creating a potential ingestion surface.
Ingestion points: assets/run-state.template.json (used to track task IDs and hashes).
Boundary markers: The skill instructions provide explicit rules for allowed state transitions defined in references/state-machine.md.
Capability inventory: Local subprocess execution of the validate-transition.py script.
Sanitization: The validation script checks all state changes against a whitelist, preventing arbitrary or illegal state transitions by returning an error status if validation fails.

agent-run-supervisor