The Agent Skills Directory

[DYNAMIC_EXECUTION]: The test suite (scripts/tests/test_checks_toy.py) uses importlib.util to dynamically load and execute the skill's logic from scripts/checks.py via computed file paths. Additionally, the script itself dynamically modifies sys.path to resolve internal dependencies. While used for testing and environment setup, these patterns represent dynamic code loading which can be abused if file paths are manipulated.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process user-supplied markdown artifacts (notices, memos, drafts) for review. This introduces a vulnerability surface where malicious instructions embedded in the artifacts could influence the agent's behavior or the output of the analysis script.
Ingestion points: Ingests markdown text from the artifact.text property in scripts/checks.py.
Boundary markers: No explicit delimiter or instruction-ignoring markers are applied to the artifact text within the processing script.
Capability inventory: The skill executes a local Python script (scripts/checks.py) and instructs the agent to delegate tasks to a critique-critic subagent tool.
Sanitization: The Python script performs basic string matching and regex parsing but does not sanitize or escape the content for safe interpolation into the agent's context.

critique-toy