feat: project-scoped instinct isolation

* feat: add project-scoped instinct isolation

* fix(continuous-learning-v2): harden instinct loading and promotion safety; sync v2.1 command docs

* fix(ci): make copilot-setup-steps a valid GitHub Actions workflow

* fix(hooks): stabilize docs warning inline JS regex parsing
This commit is contained in:
Harry Kwok
2026-03-02 04:07:13 +08:00
committed by GitHub
parent 2d3be88bb5
commit 5818e8adc7
26 changed files with 2476 additions and 713 deletions

View File

@@ -1,112 +1,36 @@
---
description: Cluster instincts into skills
description: Analyze instincts and suggest or generate evolved structures
agent: build
---
# Evolve Command
Cluster related instincts into structured skills: $ARGUMENTS
Analyze and evolve instincts in continuous-learning-v2: $ARGUMENTS
## Your Task
Analyze instincts and promote clusters to skills.
Run:
## Evolution Process
### Step 1: Analyze Instincts
Group instincts by:
- Trigger similarity
- Action patterns
- Category tags
- Confidence levels
### Step 2: Identify Clusters
```
Cluster: Error Handling
├── Instinct: Catch specific errors (0.85)
├── Instinct: Wrap errors with context (0.82)
├── Instinct: Log errors with stack trace (0.78)
└── Instinct: Return meaningful error messages (0.80)
```bash
python3 "${CLAUDE_PLUGIN_ROOT}/skills/continuous-learning-v2/scripts/instinct-cli.py" evolve $ARGUMENTS
```
### Step 3: Generate Skill
If `CLAUDE_PLUGIN_ROOT` is unavailable, use:
When cluster has:
- 3+ instincts
- Average confidence > 0.75
- Cohesive theme
Generate SKILL.md:
```markdown
# Error Handling Skill
## Overview
Patterns for robust error handling learned from session observations.
## Patterns
### 1. Catch Specific Errors
**Trigger**: When catching errors with generic catch
**Action**: Use specific error types
### 2. Wrap Errors with Context
**Trigger**: When re-throwing errors
**Action**: Add context with fmt.Errorf or Error.cause
### 3. Log with Stack Trace
**Trigger**: When logging errors
**Action**: Include stack trace for debugging
### 4. Meaningful Messages
**Trigger**: When returning errors to users
**Action**: Provide actionable error messages
```bash
python3 ~/.claude/skills/continuous-learning-v2/scripts/instinct-cli.py evolve $ARGUMENTS
```
### Step 4: Archive Instincts
## Supported Args (v2.1)
Move evolved instincts to `archived/` with reference to skill.
- no args: analysis only
- `--generate`: also generate files under `evolved/{skills,commands,agents}`
## Evolution Report
## Behavior Notes
```
Evolution Summary
=================
Clusters Found: X
Cluster 1: Error Handling
- Instincts: 5
- Avg Confidence: 0.82
- Status: ✅ Promoted to skill
Cluster 2: Testing Patterns
- Instincts: 3
- Avg Confidence: 0.71
- Status: ⏳ Needs more confidence
Cluster 3: Git Workflow
- Instincts: 2
- Avg Confidence: 0.88
- Status: ⏳ Needs more instincts
Skills Created:
- skills/error-handling/SKILL.md
Instincts Archived: 5
Remaining Instincts: 12
```
## Thresholds
| Metric | Threshold |
|--------|-----------|
| Min instincts per cluster | 3 |
| Min average confidence | 0.75 |
| Min cluster cohesion | 0.6 |
---
**TIP**: Run `/evolve` periodically to graduate instincts to skills as confidence grows.
- Uses project + global instincts for analysis.
- Shows skill/command/agent candidates from trigger and domain clustering.
- Shows project -> global promotion candidates.
- With `--generate`, output path is:
- project context: `~/.claude/homunculus/projects/<project-id>/evolved/`
- global fallback: `~/.claude/homunculus/evolved/`

View File

@@ -1,75 +1,29 @@
---
description: View learned instincts with confidence scores
description: Show learned instincts (project + global) with confidence
agent: build
---
# Instinct Status Command
Display learned instincts and their confidence scores: $ARGUMENTS
Show instinct status from continuous-learning-v2: $ARGUMENTS
## Your Task
Read and display instincts from the continuous-learning-v2 system.
Run:
## Instinct Location
Global: `~/.claude/instincts/`
Project: `.claude/instincts/`
## Status Display
### Instinct Summary
| Category | Count | Avg Confidence |
|----------|-------|----------------|
| Coding | X | 0.XX |
| Testing | X | 0.XX |
| Security | X | 0.XX |
| Git | X | 0.XX |
### High Confidence Instincts (>0.8)
```
[trigger] → [action] (confidence: 0.XX)
```bash
python3 "${CLAUDE_PLUGIN_ROOT}/skills/continuous-learning-v2/scripts/instinct-cli.py" status
```
### Learning Progress
If `CLAUDE_PLUGIN_ROOT` is unavailable, use:
- Total instincts: X
- This session: X
- Promoted to skills: X
### Recent Instincts
Last 5 instincts learned:
1. **[timestamp]** - [trigger] → [action]
2. **[timestamp]** - [trigger] → [action]
...
## Instinct Structure
```json
{
"id": "instinct-123",
"trigger": "When I see a try-catch without specific error type",
"action": "Suggest using specific error types for better handling",
"confidence": 0.75,
"applications": 5,
"successes": 4,
"source": "session-observation",
"timestamp": "2025-01-15T10:30:00Z"
}
```bash
python3 ~/.claude/skills/continuous-learning-v2/scripts/instinct-cli.py status
```
## Confidence Calculation
## Behavior Notes
```
confidence = (successes + 1) / (applications + 2)
```
Bayesian smoothing ensures new instincts don't have extreme confidence.
---
**TIP**: Use `/evolve` to cluster related instincts into skills when confidence is high.
- Output includes both project-scoped and global instincts.
- Project instincts override global instincts when IDs conflict.
- Output is grouped by domain with confidence bars.
- This command does not support extra filters in v2.1.

View File

@@ -0,0 +1,23 @@
---
description: List registered projects and instinct counts
agent: build
---
# Projects Command
Show continuous-learning-v2 project registry and stats: $ARGUMENTS
## Your Task
Run:
```bash
python3 "${CLAUDE_PLUGIN_ROOT}/skills/continuous-learning-v2/scripts/instinct-cli.py" projects
```
If `CLAUDE_PLUGIN_ROOT` is unavailable, use:
```bash
python3 ~/.claude/skills/continuous-learning-v2/scripts/instinct-cli.py projects
```

View File

@@ -0,0 +1,23 @@
---
description: Promote project instincts to global scope
agent: build
---
# Promote Command
Promote instincts in continuous-learning-v2: $ARGUMENTS
## Your Task
Run:
```bash
python3 "${CLAUDE_PLUGIN_ROOT}/skills/continuous-learning-v2/scripts/instinct-cli.py" promote $ARGUMENTS
```
If `CLAUDE_PLUGIN_ROOT` is unavailable, use:
```bash
python3 ~/.claude/skills/continuous-learning-v2/scripts/instinct-cli.py promote $ARGUMENTS
```