The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface. 1. Ingestion points: The agent is instructed to 'Explore the repo' in SKILL.md to understand the codebase. 2. Boundary markers: The instructions lack delimiters to separate untrusted repository content from the agent's system directives. 3. Capability inventory: The agent has the capability to 'publish it to the project issue tracker' as described in SKILL.md. 4. Sanitization: No explicit sanitization or escaping of ingested codebase data is required before synthesis.
[COMMAND_EXECUTION]: The skill references platform-specific commands '/setup-witify-skills' and '/to-tickets' for project environment setup and ticket creation.
[DATA_EXFILTRATION]: The skill is designed to synthesize and transmit project-related data and conversation context to an external issue tracker as its primary purpose.

to-spec