The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from task descriptions and acceptance criteria, interpolating it into prompts for subagents that have file-writing and command-execution capabilities. Evidence Chain: 1. Ingestion points: SKILL.md and references/dispatch-preparation.md. 2. Boundary markers: Absent. 3. Capability inventory: Subagents can modify files, run git commands, and execute tests as documented in references/approval-policy.md. 4. Sanitization: Absent.
[COMMAND_EXECUTION]: The skill executes various git commands such as git worktree, git merge, and git commit, alongside project-specific CLI tools to manage the development environment and session state.
[EXTERNAL_DOWNLOADS]: The skill uses npx @stevestomp/ohno-cli throughout its workflow, including the initialization and task management steps described in SKILL.md and references/ohno-integration.md. This downloads and executes a package from the NPM registry that is associated with the author's ecosystem.

work-session