The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run in lib/workflow_state/parity.py to call npx skills ls -g --json for discovering installed utilities. This execution depends on the local shell environment and the availability of Node.js.
[REMOTE_CODE_EXECUTION]: The use of npx in lib/workflow_state/parity.py can potentially trigger automatic downloads and execution of packages from the npm registry if the tool is not present in the local cache, introducing code from an external source.
[EXTERNAL_DOWNLOADS]: The npx command used for parity checking creates a network dependency on the npm registry, allowing the environment to fetch and install software during runtime.
[PROMPT_INJECTION]: The skill processes untrusted repository data such as registries and markdown tables, which could be used for indirect prompt injection.
Ingestion points: Metadata JSON files, registry.json, and slice-traceability.md are loaded in lib/workflow_state/inventory.py.
Boundary markers: No explicit delimiters or boundary markers are used to isolate user-controlled data from agent instructions.
Capability inventory: The skill has access to shell execution (subprocess.run) and Python module loading (importlib).
Sanitization: The skill uses standard regex and JSON parsing but lacks natural language sanitization to prevent the interpretation of embedded instructions.

audit-artifacts