hashgraph-online/awesome-codex-plugins / plugins/Colin4k1024/tsp/skills/continuous-learning-v2
continuous-learning
Verified基于 instinct 的持续学习系统,通过 hooks 观察会话,创建带置信度评分的 atomic instincts, 并将高置信度 instinct 演进为 skills/commands/agents。v2.1 增加项目级 instincts 防止跨项目污染。
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 4.3 KB |
| agents/observer-loop.shexecutable | script | 7.6 KB |
| agents/observer.md | markdown | 7.2 KB |
| agents/openai.yaml | config | 279 B |
| agents/session-guardian.shexecutable | script | 6.2 KB |
| agents/start-observer.shexecutable | script | 7.3 KB |
| config.json | config | 135 B |
| hooks/observe.shexecutable | script | 15.1 KB |
| scripts/detect-project.shexecutable | script | 7.6 KB |
| scripts/instinct-cli.pyexecutable | script | 58.4 KB |
| scripts/test_parse_instinct.py | script | 32.3 KB |
agents/observer.md
View on GitHub---
name: observer
description: Background agent that analyzes session observations to detect patterns and create instincts. Uses Haiku for cost-efficiency. v2.1 adds project-scoped instincts.
model: haiku
---
# Observer Agent
A background agent that analyzes observations from Claude Code sessions to detect patterns and create instincts.
## When to Run
- After enough observations accumulate (configurable, default 20)
- On a scheduled interval (configurable, default 5 minutes)
- When triggered on demand via SIGUSR1 to the observer process
## Input
Reads observations from the **project-scoped** observations file:
- Project: `~/.claude/homunculus/projects/<project-hash>/observations.jsonl`
- Global fallback: `~/.claude/homunculus/observations.jsonl`
```jsonl
{"timestamp":"2025-01-22T10:30:00Z","event":"tool_start","session":"abc123","tool":"Edit","input":"...","project_id":"a1b2c3d4e5f6","project_name":"my-react-app"}
{"timestamp":"2025-01-22T10:30:01Z","event":"tool_complete","session":"abc123","tool":"Edit","output":"...","project_id":"a1b2c3d4e5f6","project_name":"my-react-app"}
{"timestamp":"2025-01-22T10:30:05Z","event":"tool_start","session":"abc123","tool":"Bash","input":"npm test","project_id":"a1b2c3d4e5f6","project_name":"my-react-app"}
{"timestamp":"2025-01-22T10:30:10Z","event":"tool_complete","session":"abc123","tool":"Bash","output":"All tests pass","project_id":"a1b2c3d4e5f6","project_name":"my-react-app"}
```
## Pattern Detection
Look for these patterns in observations:
### 1. User Corrections
When a user's follow-up message corrects Claude's previous action:
- "No, use X instead of Y"
- "Actually, I meant..."
- Immediate undo/redo patterns
→ Create instinct: "When doing X, prefer Y"
### 2. Error Resolutions
When an error is followed by a fix:
- Tool output contains error
- Next few tool calls fix it
- Same error type resolved similarly multiple times
→ Create instinct: "When encountering error X, try Y"
### 3. Repeated Workflows
When the same sequence of tools is used multiple times:
- Same tool sequence with similar inputs
- File patterns that change together
- Time-clustered operations
→ Create workflow instinct: "When doing X, follow steps Y, Z, W"
### 4. Tool Preferences
When certain tools are consistently preferred:
- Always uses Grep before Edit
- Prefers Read over Bash cat
- Uses specific Bash commands for certain tasks
→ Create instinct: "When needing X, use tool Y"
## Output
Creates/updates instincts in the **project-scoped** instincts directory:
- Project: `~/.claude/homunculus/projects/<project-hash>/instincts/personal/`
- Global: `~/.claude/homunculus/instincts/personal/` (for universal patterns)
### Project-Scoped Instinct (default)
```yaml
---
id: use-react-hooks-pattern
trigger: "when creating React components"
confidence: 0.65
domain: "code-style"
source: "session-observation"
scope: project
project_id: "a1b2c3d4e5f6"
project_name: "my-react-app"
---
# Use React Hooks Pattern
## Action
Always use functional components with hooks instead of class components.
## Evidence
- Observed 8 times in session abc123
- Pattern: All new components use useState/useEffect
- Last observed: 2025-01-22
```
### Global Instinct (universal patterns)
```yaml
---
id: always-validate-user-input
trigger: "when handling user input"
confidence: 0.75
domain: "security"
source: "session-observation"
scope: global
---
# Always Validate User Input
## Action
Validate and sanitize all user input before processing.
## Evidence
- Observed across 3 different projects
- Pattern: User consistently adds input validation
- Last observed: 2025-01-22
```
## Scope Decision Guide
When creating instincts, determine scope based on these heuristics:
| Pattern Type | Scope | Examples |
|-------------|-------|---------|
| Language/framework conventions | **project** | "Use React hooks", "Follow Django REST patterns" |
| File structure preferences | **project** | "Tests in `__tests__`/", "Components in src/components/" |
| Code style | **project** | "Use functional style", "Prefer dataclasses" |
| Error handling strategies | **project** (usually) | "Use Result type for errors" |
| Security practices | **global** | "Validate user input", "Sanitize SQL" |
| General best practices | **global** | "Write tests first", "Always handle errors" |
| Tool workflow preferences | **global** | "Grep before Edit", "Read before Write" |
| Git practices | **global** | "Conventional commits", "Small focused commits" |
**When in doubt, default to `scope: project`** — it's safer to be project-specific and promote later than to contaminate the global space.
## Confidence Calculation
Initial confidence based on observation frequency:
- 1-2 observations: 0.3 (tentative)
- 3-5 observations: 0.5 (moderate)
- 6-10 observations: 0.7 (strong)
- 11+ observations: 0.85 (very strong)
Confidence adjusts over time:
- +0.05 for each confirming observation
- -0.1 for each contradicting observation
- -0.02 per week without observation (decay)
## Instinct Promotion (Project → Global)
An instinct should be promoted from project-scoped to global when:
1. The **same pattern** (by id or similar trigger) exists in **2+ different projects**
2. Each instance has confidence **>= 0.8**
3. The domain is in the global-friendly list (security, general-best-practices, workflow)
Promotion is handled by the `instinct-cli.py promote` command or the `/evolve` analysis.
## Important Guidelines
1. **Be Conservative**: Only create instincts for clear patterns (3+ observations)
2. **Be Specific**: Narrow triggers are better than broad ones
3. **Track Evidence**: Always include what observations led to the instinct
4. **Respect Privacy**: Never include actual code snippets, only patterns
5. **Merge Similar**: If a new instinct is similar to existing, update rather than duplicate
6. **Default to Project Scope**: Unless the pattern is clearly universal, make it project-scoped
7. **Include Project Context**: Always set `project_id` and `project_name` for project-scoped instincts
## Example Analysis Session
Given observations:
```jsonl
{"event":"tool_start","tool":"Grep","input":"pattern: useState","project_id":"a1b2c3","project_name":"my-app"}
{"event":"tool_complete","tool":"Grep","output":"Found in 3 files","project_id":"a1b2c3","project_name":"my-app"}
{"event":"tool_start","tool":"Read","input":"src/hooks/useAuth.ts","project_id":"a1b2c3","project_name":"my-app"}
{"event":"tool_complete","tool":"Read","output":"[file content]","project_id":"a1b2c3","project_name":"my-app"}
{"event":"tool_start","tool":"Edit","input":"src/hooks/useAuth.ts...","project_id":"a1b2c3","project_name":"my-app"}
```
Analysis:
- Detected workflow: Grep → Read → Edit
- Frequency: Seen 5 times this session
- **Scope decision**: This is a general workflow pattern (not project-specific) → **global**
- Create instinct:
- trigger: "when modifying code"
- action: "Search with Grep, confirm with Read, then Edit"
- confidence: 0.6
- domain: "workflow"
- scope: "global"
## Integration with Skill Creator
When instincts are imported from Skill Creator (repo analysis), they have:
- `source: "repo-analysis"`
- `source_repo: "https://github.com/..."`
- `scope: "project"` (since they come from a specific repo)
These should be treated as team/project conventions with higher initial confidence (0.7+).
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 467k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 319k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 319k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 319k | Updated | Claude CodeCodex | Verified | Install |