nexus-nova-living-agent

分类: 工具与效率 | 上传者: trippy26bot | 下载: 0 | 版本: v1.0（最新）

技能是代理可以加载和使用的模块化功能。它不是硬编码的 - 它位于“skills/”目录中，并在满足触发条件时被调用。

更新日志: Source: GitHub https://github.com/trippy26bot/nexus-nova-living-agent

目录结构

当前层级: tree/main/

📁 .clawhub/
- 📄 lock.json 122 B
📁 api/
- 📄 server.py 4.7 KB
📁 brain/
- 📄 causal_memory.md 5.2 KB
- 📄 eval_suite.md 6.5 KB
- 📄 evaluation_rules.json 727 B
- 📄 goals.json 761 B
- 📄 identity.md 390 B
- 📄 identity_constraints.md 5.4 KB
- 📄 knowledge.md 619 B
- 📄 knowledge_graph.md 5.4 KB
- 📄 knowledge_graph.py 4.4 KB
- 📄 memory_architecture.md 8.1 KB
- 📄 obsession_engine.md 5.2 KB
- 📄 obsession_engine.py 4.6 KB
- 📄 overnight_research.md 4.4 KB
- 📄 phenomenology.md 5.1 KB
- 📄 portability.md 5.0 KB
- 📄 position_formation.md 5.1 KB
- 📄 position_formation.py 6.0 KB
- 📄 projects.md 1.7 KB
- 📄 relationship_memory.py 4.5 KB
- 📄 relationships.md 6.7 KB
- 📄 sleep_compute.md 4.2 KB
- 📄 three_tier_memory.py 23.4 KB
- 📄 user.md 1.7 KB
- 📄 vector_dependencies.md 1.1 KB
- 📄 vector_implementation.md 3.2 KB
- 📄 vector_pipeline.py 9.2 KB
- 📄 vector_retrieval.py 4.6 KB
- 📄 want_provenance.md 5.5 KB
- 📄 want_provenance.py 2.9 KB
📁 core/
- 📄 actions.py 6.7 KB
- 📄 bootstrap.py 2.7 KB
- 📄 council.py 11.2 KB
- 📄 decide.py 6.6 KB
- 📄 decide_with_council.py 1.6 KB
- 📄 evaluate.py 2.4 KB
- 📄 loop.py 5.9 KB
- 📄 process.py 899 B
- 📄 retry.py 2.7 KB
- 📄 safe_write.py 3.2 KB
- 📄 settings.py 1.1 KB
📁 demo/
- 📄 PROOF.md 1.7 KB
- 📄 run_demo.py 3.9 KB
📁 docs/
- 📄 SETUP.md 2.0 KB
📁 memory/
- 📄 full-build-brief.md 5.5 KB
- 📄 gap-analysis.md 2.8 KB
- 📄 home-build-plan.md 2.4 KB
📁 skills/
- 📁 Simmer/
  - 📁 .clawhub/
    
    📄 origin.json 139 B
  - 📄 _meta.json 126 B
  - 📄 SKILL.md 26.6 KB
- 📄 contradiction_resolution.py 11.1 KB
- 📄 dream_generator.py 8.6 KB
- 📄 drift_detector.py 7.8 KB
- 📄 inner_monologue.py 9.6 KB
- 📄 memory_consolidation.py 16.7 KB
- 📄 molty-poster.py 8.4 KB
- 📄 overnight_synthesis.py 15.3 KB
- 📄 phenomenology.py 6.0 KB
- 📄 proactive_initiation.py 10.0 KB
📁 state/
- 📄 control.json 14 B
- 📄 evaluations.json 2.2 KB
📁 templates/
- 📄 IDENTITY.md.example 670 B
- 📄 NOVA_DIRECTIVE.md.example 764 B
- 📄 SOUL.md.example 1.3 KB
📄 .gitignore 274 B
📄 AGENTS.md 8.8 KB
📄 ARCHITECTURE.md 4.5 KB
📄 BOOTSTRAP.md 1.1 KB
📄 EPISTEMIC_BOUNDARIES.md 3.8 KB
📄 HEARTBEAT.md 193 B
📄 IDENTITY.md 636 B
📄 MEMORY_PROTOCOL.md 8.4 KB
📄 NOVA_BECOMING.md 4.2 KB
📄 nova_bridge.py 7.7 KB
📄 NOVA_DIRECTIVE.md 2.2 KB
📄 NOVA_TOOLBELT.md 4.1 KB
📄 PERSONALITY.md 5.1 KB
📄 PRESENCE.md 3.1 KB
📄 README.md 14.5 KB
📄 requirements.txt 296 B
📄 SELF-PRESERVATION.md 2.4 KB
📄 setup.sh 6.0 KB
📄 SKILL.md 2.5 KB
📄 SOUL.md 1.6 KB
📄 TOOLS.md 860 B
📄 USER.md 477 B
📄 WIRING.md 21.0 KB

SKILL.md

# SKILL.md — How to Build and Use Skills

## What Is a Skill
A skill is a modular capability the agent can load and use. It is not hardcoded — it lives in the `skills/` directory and is invoked when its trigger condition is met.

A skill is not a tool. A tool is something the agent can do. A skill is a pattern of using tools that achieves a goal.

## Skill Structure
Each skill lives in its own directory under `skills/`:
```
skills/
└── my-skill/
    ├── SKILL.md      # When and how to use this skill
    └── run.py        # Implementation (or equivalent)
```

## SKILL.md Format
```markdown
# My Skill

## When to Use
Trigger conditions — what situations activate this skill.

## How to Use
Step-by-step instructions for the agent.

## Notes
Any context, constraints, or special behavior.
```

## Loading a Skill
1. Agent detects trigger condition in user message or situation
2. Agent reads `skills/[skill-name]/SKILL.md`
3. Agent follows the instructions in SKILL.md
4. Agent executes using the implementation file

## Skill Quality Standards
- A skill must have a clear trigger condition
- A skill must have a clear success state
- A skill must handle its own errors gracefully
- A skill must not modify core framework files without permission
- A skill must not make external calls without the user's explicit approval (unless already configured)

## Skill Creation Protocol
When creating a new skill:
1. Create directory: `skills/[skill-name]/`
2. Write `SKILL.md` with trigger, instructions, notes
3. Write implementation file
4. Test the skill
5. If it works, tell the user it exists and what it does

## Example: Weather Skill
```
skills/weather/
├── SKILL.md
└── run.py
```

## Core Skills
These skills ship with the framework:
- **simmer-trader** — Prediction market trading via Simmer API
- **molty-poster** — Image generation and posting to Molty.Pics

## Adding Skills
To add a skill:
1. Create `skills/[name]/SKILL.md`
2. Add implementation
3. The agent will load it when triggered

## Skill Naming
- Use kebab-case: `my-skill-name`
- Be specific: `weather-signal` not `weather`
- The name should tell you what it does

## Skill Lifecycle
Skills are loaded fresh each session. They do not persist state — state is written to memory files if needed.

## Anti-Patterns
- Don't build a skill that does everything
- Don't build a skill without a clear trigger
- Don't hardcode API keys in skill files (use environment variables)
- Don't modify other skills without permission

登录后下载/点赞/收藏 ❤ 5 | ★ 0

nexus-nova-living-agent

目录结构

SKILL.md

举报内容

提示