跳到正文
cccskills
免费在 GitHub 公开

agent-pipeline

Reference for how BDB structures autonomous software engineering work — the seven-node dispatcher graph (Architect, TechLead, UI/UX, Engineering, Media/EventTech, Reviewer, Shipping) that /startcycle-graph actually runs. Use when you need the high-level lifecycle framing without inventing your own process.

中文简介正在准备中。当前显示原文说明。

查看安装方法

包含的文件(1)

  • SKILL.md5.6 KB

SKILL.md(原文)

安装前可以查看交给 AI 代理的指令内容。

🔄 BDB Software Engineering Pipeline

When orchestrating or executing autonomous software development, follow the lifecycle below — but the real, runnable implementation is .agents/graph.md, invoked via /startcycle-graph (see skills/basic/startcycle-graph/SKILL.md). This file is a lifecycle summary, not a second spec — if it and graph.md ever disagree, graph.md is right.

 define        plan          build          verify + review        ship
 ┌──────┐    ┌──────┐    ┌────────────┐    ┌──────────────┐    ┌─────────┐
 │ Goal │ ─▶ │ Plan │ ─▶ │ UI/UX,     │ ─▶ │ Reviewer     │ ─▶ │ Shipping│
 │      │    │      │    │ Engineering,│    │ (adversarial)│    │ (gate)  │
 │      │    │      │    │ Media/Event│    │              │    │         │
 └──────┘    └──────┘    └────────────┘    └──────────────┘    └─────────┘
 Architect   TechLead     the build nodes    no-progress guard   requires GO
                          run in parallel     repairs loop back   to actually ship

There is no separate /spec, /plan, /build, /test, /review, or /ship slash command — an earlier version of this file described those and a Planner_Orchestrator agent that no longer exists (split into Architect + TechLead; see graph.md's Nodes section for why). The whole lifecycle above runs through the single /startcycle-graph entry point, which invokes each node in turn per graph.md's edge table:

StageNode(s)ReadsWrites
Define + PlanArchitect, TechLeadstate.goalstate.artifacts.plan
BuildGodmode_UI_UX, Godmode_Engineering, Godmode_Media_EventTech (if needed)the plantheir own artifacts
Verify + ReviewReviewerbuild artifacts + contract only, never the implementer's claimstate.findings
ShipGodmode_Shippingall artifacts, findingsstate.gate, requires a human GO

The one rule, unchanged from graph.md: these nodes never invoke each other. A dispatcher — the /startcycle-graph workflow, or the main session acting as one — reads production_artifacts/state.json after each node returns and decides what runs next.

1. Overview

This file exists so anyone (human or agent) asking "what's the BDB software lifecycle?" gets the real shape of it, without re-deriving it from graph.md line by line. It is descriptive, not an alternate entry point.

Optional post-ship step, outside the graph: /design-control-loop sets up a recurring improvement loop in CI for a quality goal. It is started manually and the dispatcher never calls it.

2. When to Use

  • Use when you need the lifecycle framing (define → plan → build → verify → ship) to structure a conversation or a manual walkthrough.
  • To actually run the pipeline autonomously, invoke /startcycle-graph — don't hand-simulate the stages here.
  • Exclude when standard tool execution is sufficient for the task at hand.

3. Core Process

  1. Read .agents/graph.md for the current node/edge contract before assuming anything about how a stage works — this file summarizes it, it does not own it.
  2. If autonomous execution is needed, invoke /startcycle-graph rather than manually working through the stages above.
  3. Verify exit codes, file modifications, or state.json contents to guarantee a stage's side effects actually happened before reporting completion.

4. Common Rationalizations

RationalizationReality
"The code change was small, so I skipped updating OpenWiki docs."Every state change must be reflected in the relevant system records.
"The ingest script exited without an error, so the memB index must be updated."Silent failures happen; explicit verification of the side effect is mandatory.
"I'll let /startcycle-graph proceed without a defined rollback path."Proceeding without a rollback path corrupts the workflow integrity and safety.
"This file's stage descriptions look complete, so I'll follow them 'in spirit' instead of invoking /startcycle-graph."Exactly the failure this file was rewritten to stop — see graph.md and SESSION-HANDOVER-v3.13.md for the incident where a duplicated pipeline description caused a model to skip the real dispatcher entirely.

5. Red Flags

  • Referencing /spec, /plan, /build, /test, /review, or /ship as if they are real slash commands in this repo — they are not.
  • Referencing Planner_Orchestrator as a current agent — it was split into Architect and TechLead.
  • One node's output instructing another node directly instead of the dispatcher deciding the next step from state.json (the hand-off pattern graph.md (F-17) rules out).
  • Bypassing the verification step after a script execution.

6. Verification

  • Verified script exit codes are explicitly checked.
  • Confirmed target files or state.json reflect the expected change.
  • Ensured no silent failures were ignored before reporting success.

评价

还没有评价。欢迎分享你的使用感受。

评论

无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。

还没有评论。来写第一条吧。

提交后会立即公开。请勿填写邮箱、电话等个人信息。URL 最多 2 个。

0 / 1000 字

此表单受 reCAPTCHA 保护,适用 Google 的隐私权政策和服务条款。

同一仓库的技能

Use when manage multiple local CLI agents via tmux sessions (start/stop/monitor/assign) with cron-friendly scheduling.

中文简介正在准备中。当前显示原文说明。

hybridlabor-api/aos62026年10月16日 更新

Use when a hybrid memory system that provides persistent, searchable knowledge management for AI agents (Architecture, Patterns, Decisions).

中文简介正在准备中。当前显示原文说明。

hybridlabor-api/aos62026年10月16日 更新

Tools are how AI agents interact with the world. A well-designed tool is the difference between an agent that works and one that hallucinates, fails silently, or costs 10x more tokens than necessary. This skill covers tool design from schema to error handling.

中文简介正在准备中。当前显示原文说明。

hybridlabor-api/aos62026年10月16日 更新

Harness patterns for coding agents — memory, permissions, context engineering, delegation, skills, hooks, bootstrap.

中文简介正在准备中。当前显示原文说明。

hybridlabor-api/aos62026年10月16日 更新

agenttrail: live map of a multi-agent build in the browser: which plan component is being worked on, by which agent or harness, what is done and what is stuck. Use when a multi-agent pipeline starts (/startcycle, /startcycle-graph, /teamwork-preview) or after a plan-canvas approve, when the user asks to see what the agents are doing ("live map", "build board", "who is running now"), or by yourself whenever a multi-agent run is under way — no user prompt needed.

中文简介正在准备中。当前显示原文说明。

hybridlabor-api/aos62026年10月16日 更新

AI agent development workflow for building autonomous agents, multi-agent systems, and agent orchestration with CrewAI, LangGraph, and custom agents.

中文简介正在准备中。当前显示原文说明。

hybridlabor-api/aos62026年10月16日 更新

查看 hybridlabor-api 的全部技能

报告此技能的问题