The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources, specifically project specifications, handoffs, and tickets, which are used to generate instructions for sub-agents and content for pull requests.
Ingestion points: Step 1 instructions to 'Read the spec, the handoff and every ticket in full' from the effort folder.
Boundary markers: No explicit delimiters or instructions are provided to the agent to treat the content of tickets as untrusted data or to ignore potential instructions embedded within them.
Capability inventory: The skill has significant capabilities including delegating tasks via /implement, performing git operations (commit, push, cherry-pick), running testing frameworks (npx playwright), and creating pull requests via /to-pr.
Sanitization: There is no evidence of sanitization or validation of the input data before it is interpolated into sub-agent briefs or pull request descriptions.
[COMMAND_EXECUTION]: The skill performs various shell-level operations as part of its orchestration logic.
Git Operations: Step 2 and Step 5 involve git commit, git push, and git cherry-pick --no-commit to manage the codebase.
Testing Tools: Step 4 specifies the use of headless browsers and testing tools, including npx playwright and the project's own internal test suites.
[EXTERNAL_DOWNLOADS]: The skill's instructions reference the use of external tools for environment management.
Environment Tools: The skill expects parameters for a <worktree-tool> (e.g., treehouse) and a <session-host> (e.g., herdr, Claude Code Desktop) to manage the execution environment and worktree lifecycle.

orchestrate-effort