The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from several project locations including decisions/, context/decisions/, docs/decisions/, and adr/ directories. These serve as ingestion points where instructions embedded in processed files could potentially influence agent behavior. While the skill uses a structured Markdown template and requires human affirmation for final state changes, it lacks explicit sanitization of the content read from these files.
[COMMAND_EXECUTION]: The skill is instructed to execute a local utility script (python tools/allocate-id.py) to manage decision identifiers. This represents a functional command execution capability that depends on the integrity of the local workspace environment and the specific script provided within it.

decision-record