The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources including service definitions, incident logs, and an optional instruction overlay, which could contain malicious instructions.
Ingestion points: The skill reads services/<id>.yml, incidents/incidents.json, and files within the directory specified by the $POWERSUIT_OVERLAY environment variable (e.g., README.md).
Boundary markers: The instructions do not define boundary markers or include warnings to ignore instructions embedded within the ingested data.
Capability inventory: The skill possesses file read/write capabilities (updating services/<id>.yml) and the ability to execute shell commands and Python scripts.
Sanitization: There is no evidence of sanitization or schema validation for the ingested YAML or JSON data before it is processed or used to influence agent behavior.
[DYNAMIC_EXECUTION]: The skill executes local Python scripts to perform calculations and checks.
The skill copies build_status.py from a template directory and then executes it using python3 scripts/build_status.py --check.
It also uses python3 -c for one-liner downtime calculations, which involves dynamic evaluation of arguments.
[COMMAND_EXECUTION]: The skill performs several shell operations to manage the environment.
It uses mkdir, rsync, and ls to set up directories, copy templates, and check for the existence of overlay files.

nines-per-service