A multi-agent framework that generates, modifies, and debugs CI/CD pipelines through conversation.
389
A multi-agent framework that generates, modifies, and debugs CI/CD pipelines through conversation.
CI-Copilot scans your repository, understands your stack, and produces production-ready workflows — starting with GitHub Actions, with more platforms on the way. It's not a template engine. Specialized AI agents collaborate to detect your project context, validate against security and compliance policies, render the final pipeline, and commit it as a PR. You approve at every step.
GitHub: github.com/talkops-ai/ci-copilot
You need two things: a .env file and a docker-compose.yml.
1. Create a .env file:
OPENAI_API_KEY=your_openai_api_key
GITHUB_PERSONAL_ACCESS_TOKEN=your_github_pat_with_repo_scope
2. Create a docker-compose.yml:
services:
ci-copilot:
image: talkopsai/ci-copilot:latest
container_name: ci-copilot
ports:
- "10102:10102"
environment:
- OPENAI_API_KEY=${OPENAI_API_KEY}
- GITHUB_PERSONAL_ACCESS_TOKEN=${GITHUB_PERSONAL_ACCESS_TOKEN}
# LLM config — defaults to OpenAI. See .env.example in the repo for other providers.
- LLM_PROVIDER=openai
- LLM_MODEL=gpt-4o-mini
- LLM_HIGHER_PROVIDER=openai
- LLM_HIGHER_MODEL=gpt-5-mini
- LLM_DEEPAGENT_PROVIDER=openai
- LLM_DEEPAGENT_MODEL=o4-mini
- LOG_LEVEL=INFO
restart: unless-stopped
networks:
- ci-copilot-net
talkops-ui:
image: talkopsai/talkops-ui:latest
container_name: talkops-ui
environment:
- TALKOPS_CI_COPILOT_URL=http://ci-copilot:10102
ports:
- "8080:80"
depends_on:
- ci-copilot
restart: unless-stopped
networks:
- ci-copilot-net
networks:
ci-copilot-net:
driver: bridge
3. Start everything:
docker compose up -d
http://localhost:10102http://localhost:8080Open the UI and try: "Create a CI/CD pipeline for my-org/my-app"
actionlint, and auto-corrects via a refinement loop| Variable | Required | Default | Description |
|---|---|---|---|
OPENAI_API_KEY | Yes* | — | LLM API key (default provider is OpenAI) |
GITHUB_PERSONAL_ACCESS_TOKEN | Yes | — | GitHub PAT with repo scope |
LLM_PROVIDER | No | openai | openai, anthropic, google_genai, bedrock, azure_openai |
LLM_MODEL | No | gpt-4o-mini | Model for standard tasks |
LLM_HIGHER_PROVIDER / LLM_HIGHER_MODEL | No | openai / gpt-5-mini | Model for complex reasoning |
LLM_DEEPAGENT_PROVIDER / LLM_DEEPAGENT_MODEL | No | openai / o4-mini | Model for code generation |
LOG_LEVEL | No | INFO | DEBUG, INFO, WARNING, ERROR |
*If using a different LLM provider, supply its API key instead (e.g.
ANTHROPIC_API_KEY,GOOGLE_API_KEY).
| Port | Protocol | Description |
|---|---|---|
10102 | HTTP | A2A server (agent-to-agent protocol) |
latest — most recent stable build0.1.0 — initial releaseContent type
Image
Digest
sha256:932d48225…
Size
114.9 MB
Last updated
8 months ago
docker pull talkopsai/ci-copilot