skill-pipeline

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Skill Pipeline

技能Pipeline

The conductor, not a player. This skill classifies tasks, selects the pipeline variant, calibrates depth, and orchestrates handoffs between skills. It produces no artifacts of its own — its output is routing decisions that activate other skills.

它是指挥者，而非执行者。该技能负责对任务分类、选择Pipeline变体、校准执行深度，并且协调各个技能之间的交接。它本身不会生成任何产物——输出内容是用于激活其他技能的路由决策。

Task Classification

任务分类

On every coding task, classify before acting. Evaluate scope signals and map to a task class.

Input signals: file count, task description, existing plan/handoff files, batch indicators, CI environment.

Task received
  │
  ├─ Trivial (typo, rename, version bump)
  │  → No skills. Just do it.
  │
  ├─ Small (isolated fix, single-file, <10 logic lines)
  │  → simplify-and-harden only (post-completion)
  │
  ├─ Medium (feature in known area, 2-5 files)
  │  → intent-framed-agent + simplify-and-harden
  │
  ├─ Large (complex refactor, new architecture, unfamiliar codebase, high-risk logic)
  │  → Full standard pipeline
  │  → Recommend /plan-interview before starting
  │
  ├─ Long-running (multi-session, high context pressure, prior handoff exists)
  │  → Full standard pipeline with context-surfing as critical skill
  │
  └─ Batch (multiple features from spec, 5+ discrete tasks, issue triage)
     → Team-based pipeline (agent-teams-simplify-and-harden)

When uncertain, start with Medium. Add skills if drift or quality issues appear mid-task.

For detailed heuristics, edge cases, and examples: read

references/classification-rules.md

处理任何编码任务时，都要先分类再执行。评估范围信号并映射到对应的任务类别。

输入信号： 文件数量、任务描述、已有的计划/交接文件、批量标识、CI环境。

Task received
  │
  ├─ Trivial (typo, rename, version bump)
  │  → No skills. Just do it.
  │
  ├─ Small (isolated fix, single-file, <10 logic lines)
  │  → simplify-and-harden only (post-completion)
  │
  ├─ Medium (feature in known area, 2-5 files)
  │  → intent-framed-agent + simplify-and-harden
  │
  ├─ Large (complex refactor, new architecture, unfamiliar codebase, high-risk logic)
  │  → Full standard pipeline
  │  → Recommend /plan-interview before starting
  │
  ├─ Long-running (multi-session, high context pressure, prior handoff exists)
  │  → Full standard pipeline with context-surfing as critical skill
  │
  └─ Batch (multiple features from spec, 5+ discrete tasks, issue triage)
     → Team-based pipeline (agent-teams-simplify-and-harden)

如果无法确定分类，先按中等任务处理。如果任务执行过程中出现范围偏移或质量问题，可补充对应技能。

如需了解详细的启发式规则、边界场景和示例，请阅读

references/classification-rules.md

。

Pipeline Selection

Pipeline选择

Route task class to the right variant:

Task Class	Variant	Rationale
Trivial	None	No overhead needed
Small	Standard (minimal)	S&H only
Medium	Standard (partial)	Scope monitoring + review
Large	Standard (full)	Full depth with planning
Long-running	Standard (full)	Context-surfing is critical
Batch	Team-based	Breadth over depth
CI environment	CI	Headless review

Heuristic: Standard pipeline for depth (single complex feature). Team-based pipeline for breadth (batch of tasks). CI pipeline when

CI=true

GITHUB_ACTIONS=true

将任务类别匹配到合适的变体：

任务类别	变体	设计逻辑
琐碎任务	无	不需要额外开销
小任务	标准型（最小配置）	仅启用S&H
中等任务	标准型（部分配置）	范围监控 + 评审
大型任务	标准型（全量配置）	带规划的全深度执行
长周期任务	标准型（全量配置）	context-surfing为核心能力
批量任务	团队协作型	优先覆盖广度而非深度
CI环境	CI型	无界面评审

启发式规则： 标准Pipeline适用于要求「深度」的场景（单个复杂功能），团队协作型Pipeline适用于要求「广度」的场景（批量任务），当环境变量

CI=true

或

GITHUB_ACTIONS=true

时使用CI Pipeline。

Activation Sequences

激活顺序

Standard Pipeline

标准Pipeline

classify
  → (recommend /plan-interview if Large or Long-running)
  → intent-framed-agent (at planning-to-execution transition)
  → context-surfing (auto-activates when intent frame + plan exist; concurrent with intent monitoring)
  → [IMPLEMENTATION]
  → simplify-and-harden (post-completion, if non-trivial diff)
  → self-improvement (on errors, corrections, or S&H learning candidates)

Skill-by-class activation:

Skill	Trivial	Small	Medium	Large	Long-running
plan-interview	-	-	-	Recommend	Recommend
intent-framed-agent	-	-	Activate	Activate	Activate
context-surfing	-	-	-	Activate	Critical
simplify-and-harden	-	If non-trivial	If non-trivial	If non-trivial	If non-trivial
self-improvement	On error only	On error only	On error/completion	On error/completion	On error/completion

classify
  → (recommend /plan-interview if Large or Long-running)
  → intent-framed-agent (at planning-to-execution transition)
  → context-surfing (auto-activates when intent frame + plan exist; concurrent with intent monitoring)
  → [IMPLEMENTATION]
  → simplify-and-harden (post-completion, if non-trivial diff)
  → self-improvement (on errors, corrections, or S&H learning candidates)

按任务类别的技能激活规则：

技能	琐碎任务	小任务	中等任务	大型任务	长周期任务
plan-interview	-	-	-	推荐	推荐
intent-framed-agent	-	-	激活	激活	激活
context-surfing	-	-	-	激活	核心必备
simplify-and-harden	-	非平凡变更时启用	非平凡变更时启用	非平凡变更时启用	非平凡变更时启用
self-improvement	仅出错时启用	仅出错时启用	出错/完成时启用	出错/完成时启用	出错/完成时启用

Team-Based Pipeline

团队协作型Pipeline

classify (Batch)
  → (recommend /plan-interview if no spec exists)
  → agent-teams-simplify-and-harden
    ├─ Team lead emits Intent Frame #1
    ├─ Phase 1: parallel implementation agents
    ├─ Compile + test verification
    ├─ Phase 2: parallel audit agents (simplify, harden, spec)
    ├─ Fix loop (up to 3 audit rounds)
    └─ Learning loop output
  → self-improvement

classify (Batch)
  → (recommend /plan-interview if no spec exists)
  → agent-teams-simplify-and-harden
    ├─ Team lead emits Intent Frame #1
    ├─ Phase 1: parallel implementation agents
    ├─ Compile + test verification
    ├─ Phase 2: parallel audit agents (simplify, harden, spec)
    ├─ Fix loop (up to 3 audit rounds)
    └─ Learning loop output
  → self-improvement

CI Pipeline

classify (CI detected)
  → simplify-and-harden-ci (headless scan, PR changed files only)
  → self-improvement-ci (pattern aggregation, promotion recommendations)

classify (CI detected)
  → simplify-and-harden-ci (headless scan, PR changed files only)
  → self-improvement-ci (pattern aggregation, promotion recommendations)

Depth Calibration

深度校准

Not just which skills — how deep each goes:

Dimension	Small	Medium	Large	Long-running	Batch
Planning passes	0	0-1	1-2	Deep iterative	Per-task or umbrella
Intent frame	-	Single frame	Full frame + monitoring	Full + handoff	Team lead frame
Context-surfing	-	-	Active	Critical (exit protocol ready)	Lightweight drift checks
S&H budget	20% diff, 60s	20% diff, 60s	20% diff, 60s	20% diff, 60s	30% team growth cap
Audit rounds (teams)	-	-	-	-	Up to 3
Self-improvement	Error-triggered	Error-triggered	Error + S&H feed	Error + S&H feed	Error + teams feed

不仅要确定使用哪些技能，还要校准每个技能的执行深度：

维度	小任务	中等任务	大型任务	长周期任务	批量任务
规划迭代次数	0	0-1	1-2	深度迭代	按任务或统一规划
意图框架	-	单框架	完整框架 + 监控	完整框架 + 交接	团队负责人输出框架
context-surfing	-	-	活跃	核心（已准备退出协议）	轻量偏移检查
S&H预算	20% diff, 60s	20% diff, 60s	20% diff, 60s	20% diff, 60s	30%团队成长上限
（团队）审核轮次	-	-	-	-	最多3轮
self-improvement	错误触发	错误触发	错误 + S&H反馈	错误 + S&H反馈	错误 + 团队反馈

Handoff Rules

交接规则

Artifacts flow between skills. The orchestrator ensures each skill receives what it needs.

Key handoffs:

Plan file (

docs/plans/plan-NNN-<slug>.md

) — produced by

plan-interview

, consumed by

intent-framed-agent

(context),

context-surfing

(wave anchor),

agent-teams

(task extraction).

Intent Frame — produced by
```
intent-framed-agent
```
, consumed by
```
context-surfing
```
(wave anchor strengthening). Copied into handoff files on drift exit.
Handoff file (
```
.context-surfing/handoff-[slug]-[timestamp].md
```
) — produced by
```
context-surfing
```
on drift exit, consumed by next session for resume.
Learning candidates (
```
learning_loop.candidates
```
) — produced by
```
simplify-and-harden
```
and
```
agent-teams
```
, consumed by
```
self-improvement
```
for pattern tracking.

Precedence: If

context-surfing

and

intent-framed-agent

both fire simultaneously, context-surfing's exit takes precedence. Degraded context makes scope checks unreliable.

For the full artifact/signal/budget table: read

references/handoff-matrix.md

产物在技能之间流转，编排器会确保每个技能收到所需的输入。

核心交接项：

计划文件 (
```
docs/plans/plan-NNN-<slug>.md
```
) — 由
```
plan-interview
```
生成，供
```
intent-framed-agent
```
（上下文）、
```
context-surfing
```
（波动锚点）、
```
agent-teams
```
（任务提取）使用。
意图框架 — 由
```
intent-framed-agent
```
生成，供
```
context-surfing
```
（强化波动锚点）使用。发生范围偏移退出时会复制到交接文件中。
交接文件 (
```
.context-surfing/handoff-[slug]-[timestamp].md
```
) — 发生范围偏移退出时由
```
context-surfing
```
生成，供下一次会话恢复上下文使用。

学习候选集 (

learning_loop.candidates

) — 由

simplify-and-harden

和

agent-teams

生成，供

self-improvement

做模式追踪使用。

优先级规则： 如果

context-surfing

和

intent-framed-agent

同时触发，context-surfing的退出请求优先级更高。上下文降级会导致范围检查不可靠。

如需查看完整的产物/信号/预算表，请阅读

references/handoff-matrix.md

。

Decision Points

决策节点

The orchestrator intervenes at these moments:

编排器会在以下节点介入：

Task Arrival

任务接收时

Classify the task. Select pipeline variant and depth. Emit routing decision. If Large/Long-running, recommend

/plan-interview

. If Batch, recommend team-based variant.

对任务分类，选择Pipeline变体和深度，输出路由决策。如果是大型/长周期任务，推荐使用

/plan-interview

。如果是批量任务，推荐使用团队协作型变体。

Plan Approval

计划获批时

When user approves a plan from

plan-interview

, flow directly into the execution stage — no separate "should I proceed?" prompt. This means activating

intent-framed-agent

to emit an Intent Frame. The intent frame itself still requires user confirmation before coding begins (that confirmation is part of

intent-framed-agent

, not an extra gate). Populate task tracking with checklist items.

当用户批准

plan-interview

输出的计划后，直接进入执行阶段——不需要额外的「是否继续」确认。这意味着会激活

intent-framed-agent

生成意图框架，编码开始前仍然需要用户确认意图框架（该确认是

intent-framed-agent

的内置流程，不是额外关卡）。同时填充任务跟踪的检查清单项。

Planning-to-Execution Transition

规划转执行阶段

When no plan-interview was used and the user signals readiness ("go ahead", "implement this", "let's start"), activate

intent-framed-agent

. Emit Intent Frame. Wait for user confirmation of the frame before coding.

如果没有使用plan-interview，且用户发出就绪信号（「继续」、「实现这个需求」、「开始吧」），则激活

intent-framed-agent

生成意图框架，等待用户确认框架后再开始编码。

Implementation Complete

实现完成时

Check if diff meets the non-trivial threshold (see

references/classification-rules.md

). If yes, activate

simplify-and-harden

. If no, signal completion directly.

检查diff是否达到非平凡变更阈值（参考

references/classification-rules.md

）。如果是，激活

simplify-and-harden

；如果否，直接发出完成信号。

Drift Detected

检测到范围偏移时

context-surfing

fires a drift exit, stop execution. Write handoff file. If the task was classified below Large, consider re-classifying upward for the next session.

如果

context-surfing

触发范围偏移退出，停止执行，写入交接文件。如果原任务分类低于大型任务，可考虑在下一次会话升级分类。

Session Resume

会话恢复时

Check for handoff files in

.context-surfing/

. If found, read completely. Re-establish context from handoff. Re-classify if needed. Resume from recommended re-entry point.

检查

.context-surfing/

目录下是否有交接文件。如果有，完整读取文件内容，从交接信息中重建上下文，必要时重新分类，从推荐的重入节点恢复执行。

Overrides

手动覆盖规则

Users can override any routing decision:

Force depth:
```
depth=small
```
/
```
depth=large
```
— override classification
Force variant:
```
variant=teams
```
/
```
variant=standard
```
— override pipeline selection
Skip review:
```
--no-review
```
— skip
```
simplify-and-harden
```
Force planning: invoke
```
/plan-interview
```
on any task regardless of classification
Skip all skills: user says "just do it" on a non-trivial task — respect the override

用户可以覆盖任何路由决策：

强制指定深度：
```
depth=small
```
/
```
depth=large
```
— 覆盖分类结果
强制指定变体：
```
variant=teams
```
/
```
variant=standard
```
— 覆盖Pipeline选择结果
跳过评审：
```
--no-review
```
— 跳过
```
simplify-and-harden
```
步骤
强制规划： 任意任务都可以调用
```
/plan-interview
```
，不受分类限制
跳过所有技能： 用户对非平凡任务下达「直接执行」指令时，尊重该覆盖规则

Re-classification

重分类规则

Tasks can change class mid-execution. Watch for:

Escalation signals: scope expanded beyond original estimate, many more files affected than expected,
```
context-surfing
```
drift exit,
```
intent-framed-agent
```
detects significant scope change
De-escalation signals: task turns out simpler than planned, plan reveals minimal changes needed

When re-classification is warranted:

Note the signal that triggered re-classification
Adjust active skills (add or remove pipeline stages)
If escalating to Large: recommend
```
/plan-interview
```
if no plan exists
If de-escalating: drop unnecessary stages, proceed with lighter pipeline

任务执行过程中可能变更类别，需要关注以下信号：

升级信号： 范围超出初始预估、受影响文件远超预期、
```
context-surfing
```
触发范围偏移退出、
```
intent-framed-agent
```
检测到重大范围变更
降级信号： 任务实际复杂度低于预期、规划显示只需要极小改动

当需要重分类时：

记录触发重分类的信号
调整活跃技能（新增或移除Pipeline阶段）
如果升级为大型任务且无现成计划，推荐使用
```
/plan-interview
```
如果降级，移除不必要的阶段，使用更轻量的Pipeline继续执行

Anti-Patterns

反模式

Do NOT re-implement skill logic. This skill classifies and dispatches. Each individual skill owns its own procedure.
Do NOT auto-invoke plan-interview. It is a human gate. Recommend it; let the user decide.
Do NOT create a third monitoring layer. During execution,
```
intent-framed-agent
```
monitors scope and
```
context-surfing
```
monitors quality. The orchestrator dispatches at decision points, then gets out of the way.
Do NOT override individual skill guardrails. If
```
simplify-and-harden
```
has a 20% budget cap, the orchestrator respects it.
Do NOT force skills on trivial tasks. The pipeline exists to help, not to slow down simple work.

不要重复实现技能逻辑： 该技能仅负责分类和调度，每个独立技能自行管理内部流程
不要自动调用plan-interview： 它是人工管控节点，仅做推荐，由用户决定是否使用
不要创建第三层监控： 执行阶段由
```
intent-framed-agent
```
监控范围，
```
context-surfing
```
监控质量，编排器仅在决策节点调度，之后不介入执行
不要覆盖单个技能的防护规则： 如果
```
simplify-and-harden
```
设置了20%的预算上限，编排器需要遵守该限制
不要强制给琐碎任务挂载技能： Pipeline的作用是提效，而非拖慢简单任务的执行速度

Pipeline Variant Details

Pipeline变体详情

For complete step-by-step walkthroughs of each variant including hybrid scenarios and session resume: read

references/pipeline-variants.md

如需查看每个变体的完整分步流程，包括混合场景和会话恢复的说明，请阅读

references/pipeline-variants.md

。