The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains instructions that explicitly direct sub-agents to ignore the 'Collaboration Protocol' requirement for manual user approval before writing files. This 'bounded exception' allows sub-agents to write directly to specific paths in the production/, docs/, and tests/ directories, effectively overriding platform-level consistency guidelines for user confirmation.
[DYNAMIC_CONTEXT_INJECTION]: The skill uses the dynamic injection syntax (!) to execute a shell command at load time: !bash "${CLAUDE_SKILL_DIR}/../../hooks/yaml-helper.sh" resolve_config --keys review_mode,automation,team.size. This invokes a local utility script to populate configuration variables.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from untrusted local files to drive its logic, specifically reading from production/session-state/active.md and milestone files to infer the release version.
Ingestion points: Version inference logic in SKILL.md reads milestone and session state files.
Boundary markers: Employs AskUserQuestion to confirm the inferred version with the user before proceeding.
Capability inventory: Spawns sub-agents with Write, Edit, and Bash capabilities across the release pipeline.
Sanitization: No explicit sanitization or validation logic is present for the version string inferred from the project files.

team-release