skill-exporter

⚠Review·Scanned 2/18/2026

Exports Clawdbot skills into standalone microservices and generates Docker/FastAPI/LLM integration. The workflow and generated code instruct running python3 {baseDir}/scripts/export.py and docker-compose up, read ANTHROPIC_API_KEY/OPENAI_API_KEY, and make LLM API calls.

from clawhub.ai·v1.0.0·18.7 KB·0 installs

Scanned from 1.0.0 at befc0f4 · Transparency log ↗

$ vett add clawhub.ai/macstenk/skill-exporterReview findings below

Skill Exporter

Transform Clawdbot skills into standalone, deployable microservices.

Workflow

Clawdbot Skill (tested & working)
         ↓
    skill-exporter
         ↓
Standalone Microservice
         ↓
Railway / Fly.io / Docker

Usage

Export a skill

python3 {baseDir}/scripts/export.py \
  --skill ~/.clawdbot/skills/instagram \
  --target railway \
  --llm anthropic \
  --output ~/projects/instagram-service

Options

Flag	Description	Default
`--skill`	Path to skill directory	required
`--target`	Deployment target: `railway`, `fly`, `docker`	`docker`
`--llm`	LLM provider: `anthropic`, `openai`, `none`	`none`
`--output`	Output directory	`./<skill-name>-service`
`--port`	API port	`8000`

Targets

railway — Generates railway.json, optimized Dockerfile, health checks fly — Generates fly.toml, multi-region ready docker — Generic Dockerfile, docker-compose.yml

LLM Integration

When --llm is set, generates llm_client.py with:

Caption/prompt generation
Decision making helpers
Rate limiting and error handling

What Gets Generated

<skill>-service/
├── Dockerfile
├── docker-compose.yml
├── api.py              # FastAPI wrapper
├── llm_client.py       # If --llm specified
├── requirements.txt
├── .env.example
├── railway.json        # If --target railway
├── fly.toml            # If --target fly
└── scripts/            # Copied from original skill
    └── *.py

Requirements

The source skill must have:

SKILL.md with valid frontmatter
At least one script in scripts/
Scripts should be callable (functions, not just inline code)

Post-Export

Copy .env.example to .env and fill in secrets
Test locally: docker-compose up
Deploy: railway up or fly deploy