메뉴
Protocol for agents to write structured outputs that survive session resets and are findable by the orchestrator.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Deterministic AI development workflow orchestrator. Manages task boards (.jonggrang/jonggrang-tasks.json), runs 16-phase pipelines (BUGFIX/SMALL/MEDIUM/LARGE), installs lifecycle hooks for Claude Code and OpenCode, and coordinates five specialist roles (Lead/Developer/Reviewer/TestLead/Tester). Use when planning features, executing multi-step development workflows, or orchestrating AI coding agents. Run via npx jonggrang.
16-phase feature orchestration workflow. Runs in the main thread as Kernel Mode. Coordinates the Lead→Developer→Reviewer→TestLead→Tester assembly line.
GitHub workflow automation with the gh CLI — pull requests, issues, releases, the REST/GraphQL API, and safe message escaping. Auth via GH_TOKEN (set in global settings, attached to the sandbox env).
GitLab workflow automation with the glab CLI — merge requests, issues, CI, releases, and safe message escaping. Auth via GITLAB_TOKEN (set in global settings, attached to the sandbox env).
Prevents premature exit and infinite loops. Three mechanisms: Completion Promises, Scratchpads, Loop Detection.
Deep knowledge about deploying applications to AWS (EC2, ECS, S3, CloudFront, Cloud Run).
| name | persisting-agent-outputs |
| description | Protocol for agents to write structured outputs that survive session resets and are findable by the orchestrator. |
| type | orchestrate |
| tier | core |
| trigger | save output, persist results, write report, store findings, agent output location |
Agent outputs must be findable after session reset. The orchestrator reads these to resume work and make decisions. This skill defines the exact write protocol.
All agent outputs go to:
.jonggrang/.output/features/{feature_id}/{phase}-{role}-output.json
Examples:
.jonggrang/.output/features/auth-feature-abc123/
07-lead-architecture-plan.json
08-developer-task-001.json
09-reviewer-design-check.json
11-reviewer-code-quality.json
12-test-lead-test-plan.json
13-tester-results.json
Every output file must include this header:
{
"jonggrang-output": true,
"feature_id": "auth-feature-abc123",
"phase": 8,
"role": "developer",
"task_id": "task-001",
"agent_id": "developer-auth-001",
"timestamp": "2024-01-01T00:00:00Z",
"status": "completed",
"output": {
// role-specific payload here
}
}
Lead (architecture plan):
{
"output": {
"work_type": "MEDIUM",
"tasks": [
{ "id": "task-001", "title": "...", "description": "...", "files": [...], "blocked_by": [] },
{ "id": "task-002", "title": "...", "description": "...", "files": [...], "blocked_by": ["task-001"] }
],
"tech_decisions": [...],
"risk_factors": [...]
}
}
Developer (implementation):
{
"output": {
"task_id": "task-001",
"files_modified": ["src/auth.ts"],
"files_created": ["src/auth.test.ts"],
"summary": "Implemented JWT auth with refresh tokens",
"notes": "Used bcrypt for password hashing"
}
}
Reviewer (review report):
{
"output": {
"approved": true,
"score": 8,
"violations": [],
"warnings": ["Consider adding rate limiting"],
"required_fixes": []
}
}
Tester (test results):
{
"output": {
"tests_passed": true,
"total": 24,
"passed": 24,
"failed": 0,
"coverage": 87.3,
"failed_tests": []
}
}
After phases 8 (implementation), 12 (code-quality), and 14 (testing) complete, the orchestrator automatically records which files changed via git diff. No action required from the agent — file tracking is based on actual git state, not self-reporting.
To inspect what a phase recorded: jonggrang manifest show [feature-id]
When an agent needs to find a previous phase's output:
.jonggrang/.output/features/{feature_id}/{phase}-{role}-*.jsonThe orchestrator knows the feature_id from MANIFEST.yaml:
feature_id: auth-feature-abc123