The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple shell commands including git, npm run build, npm test, and the Gitea tea CLI for repository management and verification.
[COMMAND_EXECUTION]: The skill depends on local scripts located at ./scripts/gitea-ci-status.sh and ./scripts/gitea-pr-checks.sh to interface with Gitea's API and CI systems.
[INDIRECT_PROMPT_INJECTION]: The orchestrator ingests untrusted task data from context/backlog/ files. These files are processed to determine execution plans and worker instructions, representing a surface for malicious content to influence agent behavior.
[INDIRECT_PROMPT_INJECTION]: Evidence Chain: 1. Ingestion points: Task files located in context/backlog/. 2. Boundary markers: None identified. 3. Capability inventory: Spawning subagents via the Task tool, executing shell commands (git, npm, tea), and writing state/progress files to the .coordinator/ directory. 4. Sanitization: No explicit sanitization of task file content before interpolation into worker prompts.

gitea-coordinator