Compare commits
2 Commits
| Author | SHA1 | Date | |
|---|---|---|---|
 张宗平
|
2bba75808a | ||
|
张宗平
|
979e652f46 |
@@ -0,0 +1,170 @@
|
||||
# pi-setup
|
||||
|
||||
Pi coding agent (https://github.com/badlogic/pi-mono) — skills, extensions, and configuration for goal-driven development workflows.
|
||||
|
||||
## What This Repository Provides
|
||||
|
||||
| Component | Path | Purpose |
|
||||
|-----------|------|---------|
|
||||
| **Skill: `goal-driven-loop`** | `.agents/skills/goal-driven-loop/` | Keeps a session focused on a user-defined goal with explicit acceptance criteria. Iterates until all criteria pass or stagnation is detected. |
|
||||
| **Extension: `goal-driven-loop`** | `extensions/goal-driven-loop/` | TypeScript extension that automates trigger detection, state injection on every agent turn, context tension detection, and timed auto-default confirms. |
|
||||
| **Symlinks for Pi discovery** | `.pi/skills/`, `.pi/extensions/` | Pi-specific discovery points — point to the real source files in `.agents/skills/` and `extensions/`. |
|
||||
| **Spec & plan** | `docs/superpowers/` | Design spec and implementation plan that produced the skill + extension. |
|
||||
| **Comet archive** | `openspec/changes/archive/` | Formal change record from the OpenSpec lifecycle. |
|
||||
|
||||
The skill and extension work as a pair. The skill is the instruction layer (what the agent reads). The extension is the automation layer (mechanical trigger detection, state injection, timed confirms).
|
||||
|
||||
## Repository Layout
|
||||
|
||||
```
|
||||
pi-setup/
|
||||
├── .gitignore # ignores .claude/, .agents/, .pi/
|
||||
├── extensions/
|
||||
│ └── goal-driven-loop/ # Extension source (git-tracked)
|
||||
│ ├── index.ts # Event wiring
|
||||
│ ├── trigger-detector.ts # Phrase matching + anti-triggers
|
||||
│ ├── state-injector.ts # GOAL.md parsing + injection
|
||||
│ ├── ui-prompts.ts # Timed auto-defaults
|
||||
│ └── package.json
|
||||
├── docs/
|
||||
│ └── superpowers/
|
||||
│ ├── specs/2026-06-10-goal-driven-loop-design.md
|
||||
│ └── plans/2026-06-10-goal-driven-loop.md
|
||||
├── openspec/
|
||||
│ └── changes/archive/2026-06-10-goal-driven-loop/
|
||||
├── .agents/skills/ # (gitignored) Skill source
|
||||
│ └── goal-driven-loop/
|
||||
│ ├── SKILL.md
|
||||
│ └── references/error-handling.md
|
||||
├── .pi/ # (gitignored) Pi discovery symlinks
|
||||
│ ├── skills/goal-driven-loop → ../../.agents/skills/goal-driven-loop
|
||||
│ └── extensions/goal-driven-loop → ../../extensions/goal-driven-loop
|
||||
└── .claude/ # (gitignored) Claude Code state
|
||||
```
|
||||
|
||||
## Installation
|
||||
|
||||
The skill and extension source are committed at the top level. The agent-specific directories (`.agents/`, `.pi/`, `.claude/`) are gitignored and only hold symlinks/discovery points on the local machine.
|
||||
|
||||
### Option 1: Clone and use locally
|
||||
|
||||
```bash
|
||||
git clone <repo-url> pi-setup
|
||||
cd pi-setup
|
||||
|
||||
# Create symlinks so Pi can discover the skill and extension
|
||||
mkdir -p .agents/skills
|
||||
ln -s ../extensions/../.agents/skills/goal-driven-loop .agents/skills/goal-driven-loop
|
||||
mkdir -p .pi/skills .pi/extensions
|
||||
ln -s ../../.agents/skills/goal-driven-loop .pi/skills/goal-driven-loop
|
||||
ln -s ../../extensions/goal-driven-loop .pi/extensions/goal-driven-loop
|
||||
```
|
||||
|
||||
Or use the project's `.pi` settings to point at the source directly (see Option 3).
|
||||
|
||||
### Option 2: Install into another project (as a Git submodule)
|
||||
|
||||
```bash
|
||||
cd your-project
|
||||
git submodule add <repo-url> vendor/pi-setup
|
||||
```
|
||||
|
||||
Then add to your project's `.pi/settings.json`:
|
||||
|
||||
```json
|
||||
{
|
||||
"skills": ["vendor/pi-setup/.agents/skills"],
|
||||
"extensions": ["vendor/pi-setup/extensions/goal-driven-loop"]
|
||||
}
|
||||
```
|
||||
|
||||
### Option 3: Configure Pi to discover via settings (no copy)
|
||||
|
||||
In `.pi/settings.json` (project) or `~/.pi/agent/settings.json` (user):
|
||||
|
||||
```json
|
||||
{
|
||||
"skills": ["/path/to/pi-setup/.agents/skills"],
|
||||
"extensions": ["/path/to/pi-setup/extensions/goal-driven-loop"]
|
||||
}
|
||||
```
|
||||
|
||||
Pi loads skills from any directory in this list, so the source files at the top level are used directly — no symlinks required.
|
||||
|
||||
## Verifying Installation
|
||||
|
||||
```bash
|
||||
# 1. Skill loads
|
||||
pi -p "/skill:goal-driven-loop"
|
||||
# → "Goal-Driven Loop is active. Awaiting your goal statement."
|
||||
|
||||
# 2. Extension loads
|
||||
pi -e extensions/goal-driven-loop/index.ts -p "test"
|
||||
# → no errors, normal output
|
||||
|
||||
# 3. End-to-end trigger
|
||||
mkdir -p /tmp/test && cd /tmp/test && git init
|
||||
pi -e /path/to/pi-setup/extensions/goal-driven-loop/index.ts \
|
||||
-p "当前目标是实现 reverseString(s) 函数"
|
||||
# → Goal-driven loop activates, implements, tests, completes
|
||||
```
|
||||
|
||||
## Usage
|
||||
|
||||
Set a goal with a trigger phrase (Chinese, English, or other language with similar meaning):
|
||||
|
||||
```
|
||||
设定当前目标为给 auth 模块加 rate limiting
|
||||
当前目标是实现 camelCase 函数
|
||||
my goal is to add caching, acceptance: hit rate > 80%
|
||||
```
|
||||
|
||||
The skill auto-generates verifiable acceptance criteria if you don't provide them, persists state to `.agents/goal/GOAL.md`, and iterates until all criteria pass.
|
||||
|
||||
To abandon a goal:
|
||||
|
||||
```
|
||||
放弃
|
||||
不要了
|
||||
cancel
|
||||
```
|
||||
|
||||
To override auto-defaults within the timed window, just type your choice. After 10 seconds, the system defaults to "continue" on stale state prompts.
|
||||
|
||||
## Trigger Reference
|
||||
|
||||
| Language | Patterns |
|
||||
|----------|----------|
|
||||
| Chinese | 设定当前目标为 / 当前目标是 / 把当前目标设为 / 我的目标是 |
|
||||
| English | set current goal to / current goal is / my goal is / the goal is |
|
||||
| Other | semantically equivalent phrasings |
|
||||
|
||||
The skill does **not** activate when the phrase is inside a quote block, code block, or reported speech; when the message contains trivial-task keywords (hotfix, tweak, quick fix, 修个bug, 小改); or when the goal is ≤ 10 characters without architectural keywords.
|
||||
|
||||
## Architecture
|
||||
|
||||
```
|
||||
User: "设定当前目标为 X"
|
||||
↓
|
||||
[Extension: input event] detectTrigger(X) → validates against anti-triggers
|
||||
↓
|
||||
[Extension: before_agent_start] reads .agents/goal/GOAL.md, injects state
|
||||
↓
|
||||
[Skill: SKILL.md] instructs agent on iteration steps
|
||||
↓
|
||||
[Agent] implements, verifies, updates GOAL.md
|
||||
↓
|
||||
[Extension: before_agent_start] on next turn re-injects updated state
|
||||
↓
|
||||
[Skill: completion] deslop pass + cleanup + output summary
|
||||
```
|
||||
|
||||
State is file-based — `.agents/goal/GOAL.md` — so it survives context compaction. The extension re-injects the goal state on every agent turn.
|
||||
|
||||
## Error Handling
|
||||
|
||||
The skill handles transient failures with exponential backoff (10 retries, ~37min total coverage). Stagnation uses dual-signal detection (criteria count + progress keywords) with auto-recovery injection before user notification. See `docs/superpowers/specs/2026-06-10-goal-driven-loop-design.md` for the full design.
|
||||
|
||||
## License
|
||||
|
||||
MIT (assumed — please confirm with the repository owner).
|
||||
@@ -0,0 +1,59 @@
|
||||
---
|
||||
name: goal-driven-loop
|
||||
description: Use when the user sets a current goal with phrases like "设定当前目标为", "当前目标是", "current goal is", or "my goal is", and the task is not a trivial fix, hotfix, tweak, or quoted text
|
||||
---
|
||||
|
||||
# Goal-Driven Loop
|
||||
|
||||
Keep the current session focused on a user-defined goal with explicit acceptance criteria. Iterate until all criteria pass, or genuine stagnation is detected.
|
||||
|
||||
## When to Activate
|
||||
|
||||
**Trigger phrases (any language, semantic variants):**
|
||||
- Chinese: 设定当前目标为 / 当前目标是 / 把当前目标设为 / 我的目标是
|
||||
- English: set current goal to / current goal is / my goal is / the goal is
|
||||
- Other: semantically equivalent phrasings + concrete deliverable
|
||||
|
||||
**Do NOT activate when:**
|
||||
- Trigger phrase inside markdown quote (`>`) or code block, or same sentence as reported speech (he said / 他说)
|
||||
- Message contains trivial-task keywords: hotfix, tweak, quick fix, 修个bug, 小改
|
||||
- Goal description ≤ 10 chars without architectural keywords, or agent estimates < 5 min without architectural keywords
|
||||
- `.agents/goal/GOAL.md` exists → extension prompts: continue / replace / abandon (10s timeout → continue)
|
||||
|
||||
**Granularity:** Coarse (file count > 5, multi-module, architectural) → extension invokes `comet-open`. Fine (single concern, ≤ 5 files) → this skill. Fallback: if comet unavailable, auto-split into phases.
|
||||
|
||||
## Core Loop
|
||||
|
||||
1. **Initialize** — If `GOAL.md` exists, extension prompts with timed default. If no criteria provided, auto-generate (verifiability red line). Write `GOAL.md`.
|
||||
|
||||
2. **Execute** — Read `GOAL.md`, pick next unpassed criterion (least dependent), implement, verify, record in Progress Log.
|
||||
|
||||
3. **Verify** — Check each criterion against fresh evidence. Update checkboxes, iteration counter, timestamp.
|
||||
|
||||
4. **Loop** — All ✅ → exit. Stagnation → extension auto-injects strategy change (2x before user notification). Otherwise → step 2.
|
||||
|
||||
5. **Exit** — Deslop pass on modified files (failure doesn't block). Delete `GOAL.md`. Output summary.
|
||||
|
||||
## Acceptance Criteria Verifiability
|
||||
|
||||
Auto-generated criteria must be verifiable. Examples:
|
||||
- ❌ "Implement rate limiting" → not verifiable
|
||||
- ✅ "Rate limiter middleware mounts on auth routes, excess requests return 429" → verifiable
|
||||
- ❌ "Code quality is good" → not verifiable
|
||||
- ✅ "Related files lint clean, typecheck passes" → verifiable
|
||||
|
||||
Agent cannot unilaterally delete or skip criteria. Material modifications require logged reason; safe adjustments (typos, splitting compounds) may be auto-applied.
|
||||
|
||||
## Goal State File
|
||||
|
||||
Path: `.agents/goal/GOAL.md` — frontmatter with `goal`, `created`, `updated`, `iteration`, `status` (in-progress | blocked | complete). Body has Acceptance Criteria (checkboxes), Progress Log (per-iteration), Blocked section.
|
||||
|
||||
## Error Handling
|
||||
|
||||
See `references/error-handling.md` for retry strategy, exception classification, stagnation rules, and red flags.
|
||||
|
||||
**Context tension:** Extension detects via `ctx.getContextUsage() > 0.8`. Auto-injects "save progress now" directive into next turn.
|
||||
|
||||
**Mid-loop granularity upgrade:** If execution reveals the goal is more complex than estimated, extension auto-invokes `comet-open` with `GOAL.md` state. User is informed via UI notification, not asked for confirmation.
|
||||
|
||||
**Explicit abandon:** User says "abandon / cancel / 不要了" → extension deletes `GOAL.md`, outputs brief summary, terminates.
|
||||
@@ -0,0 +1,75 @@
|
||||
# Error Handling Reference
|
||||
|
||||
Detailed retry strategy, exception classification, and recovery rules for the `goal-driven-loop` skill.
|
||||
|
||||
## Retry Strategy (Automatic, No Human Intervention)
|
||||
|
||||
All recoverable exceptions use exponential backoff. Counter is **per-iteration** — resets at the start of each new iteration.
|
||||
|
||||
| Parameter | Value |
|
||||
|-----------|-------|
|
||||
| Max retries | 10 per iteration |
|
||||
| Initial delay | 30s |
|
||||
| Backoff factor | ×2 |
|
||||
| Max delay | 300s (5min) |
|
||||
|
||||
**Backoff sequence:**
|
||||
|
||||
| Retry | Delay | Cumulative |
|
||||
|-------|-------|------------|
|
||||
| 1 | 30s | 0.5min |
|
||||
| 2 | 60s | 1.5min |
|
||||
| 3 | 120s | 3.5min |
|
||||
| 4 | 240s | 7.5min |
|
||||
| 5 | 300s | 12.5min |
|
||||
| 6 | 300s | 17.5min |
|
||||
| 7 | 300s | 22.5min |
|
||||
| 8 | 300s | 27.5min |
|
||||
| 9 | 300s | 32.5min |
|
||||
| 10 | 300s | 37.5min |
|
||||
|
||||
**Formula:** `delay = min(30 * 2^(retry-1), 300)` seconds.
|
||||
|
||||
**Execution:** Before retry, run `bash -c "sleep <delay> && echo done"` with `timeout` parameter set to `delay + 30`. If sleep is interrupted, treat as tool failure, enter retry.
|
||||
|
||||
**After 10 retries exhausted:** `status = blocked`, write Blocked section to `GOAL.md`, extension auto-attempts alternative approach. If still blocked after auto-attempt, then notify user.
|
||||
|
||||
## Exception Classification
|
||||
|
||||
| Exception | Detection signal | Recoverable? | Response |
|
||||
|-----------|------------------|--------------|----------|
|
||||
| Tool execution failure | bash returns non-zero exit | Yes | Exponential backoff retry |
|
||||
| LLM API exception | subagent timeout/error | Yes | Exponential backoff retry |
|
||||
| Context tension | Extension detects `getContextUsage() > 0.8` | No | Extension auto-saves state, injects continuation directive |
|
||||
| User interruption | User sends new message mid-loop | No | Extension auto-pauses, responds, resumes on next trigger |
|
||||
| Stale state | `GOAL.md` exists with `status = in-progress` | No | Extension prompts: continue / replace / abandon (10s timeout → continue) |
|
||||
|
||||
## Stagnation Detection
|
||||
|
||||
Both signals must be present:
|
||||
|
||||
1. Acceptance criteria passing count unchanged for 3 consecutive iterations, AND
|
||||
2. Progress Log last 3 iterations contain no progress keywords (found / fixed / implemented / added / refactored / verified)
|
||||
|
||||
Only both conditions together trigger stagnation. Extension auto-injects strategy-change prompt. After 2 consecutive auto-recoveries with no progress, extension notifies user.
|
||||
|
||||
## Continuation Prompt Format (Fixed)
|
||||
|
||||
When context is tight, output exactly this format:
|
||||
|
||||
```
|
||||
🔄 目标状态已保存至 .agents/goal/GOAL.md
|
||||
当前进度：X/Y criteria 通过，第 N 轮迭代。
|
||||
```
|
||||
|
||||
The extension injects the full state summary on every `before_agent_start`, so the next prompt automatically re-loads the goal.
|
||||
|
||||
## Red Flags — These Do NOT Count as Completion
|
||||
|
||||
- "LLM call failed, goal should be considered complete" → retry or blocked
|
||||
- "Tests don't run but code looks fine" → must have test evidence
|
||||
- "Context full, let me summarize" → save state, extension auto-injects continuation, do not declare done
|
||||
- "Too many retries, skip this criterion" → exhausted retries → blocked, do not skip
|
||||
- "This criterion is too hard to verify, skip it" → agent may auto-rewrite, logged as material change
|
||||
- "All criteria basically met, close enough" → each criterion must have explicit pass evidence
|
||||
- "Deslop pass would take too long, skip" → deslop failure does not block, but agent should still attempt it
|
||||
Reference in New Issue
Block a user