The Agent Skills Directory

[PROMPT_INJECTION]: The skill features a 'verbatim' content preservation instruction in SKILL.md (Step 1) and questions.md (Question 13). This directs the agent to include user-supplied text exactly as provided in generated skill files. This creates an indirect prompt injection surface where malicious instructions could be embedded into generated artifacts, bypassing standard safety filtering.\n
Ingestion points: User-provided answers during the intake phase (Step 1).\n
Boundary markers: None; the skill explicitly instructs the agent to preserve wording 'verbatim'.\n
Capability inventory: File writing (Step 5), script execution (Step 6), and git commits (Step 8).\n
Sanitization: Absent; the instructions prioritize literal preservation of user input over sanitization.\n- [COMMAND_EXECUTION]: The skill executes a local shell script scripts/validate-skill.sh in Step 6 to check for repository conventions. The script is bundled with the skill and performs read-only checks such as line count validation and path verification.\n- [COMMAND_EXECUTION]: The skill performs repository management tasks using git add and git commit in Step 8. These operations are safeguarded by a requirement for explicit user confirmation before any changes are committed to the repository.

create-a-skill