How to operate, debug, and monitor the HCFullPipeline continuous execution engine
How to operate, debug, and monitor the HCFullPipeline continuous execution engine
npx skills add https://github.com/HeadyAI/Heady-Testing-83c4b580 --skill heady-pipeline-ops이 명령을 Claude Code에 복사하여 붙여넣어 스킬을 설치하세요
| name | heady-pipeline-ops |
| description | How to operate, debug, and monitor the HCFullPipeline continuous execution engine |
The HCFullPipeline is a 21-stage (fib(8)) DAG execution engine that runs continuously. It is the core autonomous execution backbone of the Heady™ platform.
src/hc_pipeline.js — DAG engine, topological sort, circuit breakers, worker poolsrc/hc_pipeline_handlers.js — 100+ registered task handlers (fail-closed)configs/hcfullpipeline.yaml — Stage definitions, tasks, dependencies, gate configsheady-manager.js — Boot sequence, continuous mode, API endpointsThe pipeline auto-starts on server boot with a φ² delay (~2.6s). Check status:
GET /api/buddy/pipeline/continuous → { running, cycleCount, gates }
POST /api/buddy/pipeline/continuous { "action": "start" }
POST /api/buddy/pipeline/continuous { "action": "stop" }
POST /api/pipeline/run → { runId, status, metrics }
GET /api/pipeline/state → { runId, status, metrics }
GET /api/pipeline/config → { name, version, stages, taskCount }
If a task throws [FAIL-CLOSED] No handler registered for task 'task_name', you need to register a handler:
// In src/hc_pipeline_handlers.js
registerTaskHandler('new_task_name', async (ctx) => {
// ctx has: runId, stageId, configs, supervisorConfig
return { status: 'completed', /* results */ };
});
The continuous pipeline checks 4 gates before each cycle:
If any gate fails, the pipeline halts with exitReason: "gate_failed".
Check the pipeline log at .heady/hc_pipeline.log and each stage's state in the run result object.
Use when the user wants to turn repositories, technical notes, strategy files, or mixed project materials into structured knowledge packs, briefings, or LLM-ready documentation. Helpful for repo-to-docs workflows, NotebookLM or knowledge-base preparation, executive briefings, and documentation hubs that work for both technical and non-technical audiences.
How to operate, monitor, and extend the Auto-Success Engine with its 13 φ-scaled categories
Master orchestrator prompt for the Heady ecosystem. Universal coding agent system prompt for Claude, GPT, Gemini, Perplexity, Windsurf, Cursor, Copilot, and any agentic coding system. Synthesizes all 60+ Heady skills into a single cohesive autonomous builder prompt. Use for any Heady development, architecture, deployment, orchestration, vector memory, MCP gateway, edge AI, bee swarm, liquid node, CSL engine, phi-math, sacred geometry, or full-system build task. Keywords: heady, maximum potential, universal prompt, coding agent, autonomous builder, liquid OS, sacred geometry, CSL, phi-math, vector memory, bee swarm, MCP gateway, edge AI, conductor, pipeline, alive software.
Full-spectrum Perplexity Enterprise Max integration — sonar search, deep research, reasoning, embeddings, and custom connector setup for the Heady™ ecosystem.
Automated audit procedures for verifying Sacred Geometry compliance across the entire codebase
Autonomous self-healing patterns for detecting and recovering from system failures