The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted course checkpoint data and incorporates it into generated exercises which are subsequently processed by a command-line tool.\n
Ingestion points: User-supplied checkpoint data is ingested in the workflow described in SKILL.md.\n
Boundary markers: No boundary markers or instructions are provided to the agent to distinguish between valid checkpoint data and malicious instructions.\n
Capability inventory: The skill utilizes python3 scripts/exercise_lint.py to process generated content (SKILL.md, Step 3).\n
Sanitization: There is no instruction to sanitize or escape the generated exercise text before passing it as a shell argument to the linter's --text parameter.\n- [COMMAND_EXECUTION]: The skill invokes a local Python script to perform linting tasks as a standard part of its operation.\n
Evidence: SKILL.md instructs the agent to run python3 scripts/exercise_lint.py to validate the generated exercise bank.\n- [METADATA_POISONING]: The skill's internal metadata contains an author name that does not match the system-provided author identifier.\n
Evidence: The YAML frontmatter in SKILL.md lists author: "awesome-skillkit", while the skill is attributed to x33834 by the platform information.

exercise-generator