The Agent Skills Directory

[PROMPT_INJECTION]: The skill implements defenses for potential injection surfaces. Ingestion points: reads from ./reports/*/state.json. Boundary markers: uses <user_input> tags for sub-agent prompts. Capability inventory: includes Bash, Write, Agent, and TaskUpdate. Sanitization: truncates input to 200 characters, strips backticks/angle brackets, and enforces regex format checks on repositories.\n- [COMMAND_EXECUTION]: The skill uses Bash for configuration resolution via jq, git, and gh. It prevents shell injection by using jq --arg and validating all repository-related arguments against a strict regex ([a-zA-Z0-9._-]+/[a-zA-Z0-9._-]+) before execution.

issues