The Agent Skills Directory

[COMMAND_EXECUTION]: The skill recommends executing bunx ccrecall sync for session analytics. This command invokes a third-party utility that is not part of the standard platform toolset and accesses internal platform configuration files located at ~/.claude/teams/. Evidence: Found in references/task-management.md under the 'Team Cleanup' section.
[REMOTE_CODE_EXECUTION]: The use of bunx to run the ccrecall package fetches and executes code from an external registry at runtime. Evidence: bunx ccrecall sync in references/task-management.md.
[INDIRECT_PROMPT_INJECTION]: The skill orchestrates sub-agents by interpolating task descriptions into prompts. This pattern creates a vulnerability surface for indirect prompt injection if descriptions contain untrusted data. Evidence Chain: (1) Ingestion points: Task descriptions in references/task-management.md and references/patterns.md. (2) Boundary markers: Absent. (3) Capability inventory: Significant, including file system manipulation (git worktree) and external execution (bunx). (4) Sanitization: Absent.
[EXTERNAL_DOWNLOADS]: The bunx utility downloads the ccrecall package from the npm registry if it is not already installed. Evidence: references/task-management.md.

orchestration