How rh-ai-quickstart/agentic-software-factory works
This repository demonstrates an agentic software factory: users submit requirements through a spec-collector UI, Scribe turns them into GitHub issues, Developer implements labeled issues as pull requests, and Review checks those changes against their originating specifications. Fixer and Rummager provide separate operational-response branches for pipeline failures and application-log errors. The README documents these product-level connections; sampled agent excerpts show orchestration and gateway use, but do not expose every integration implementation.
Main entry points: llama-stack-kube-github Llama Stack gateway · server.js Spec collector · app.js Collector UI · main.py Scribe agent
Components 17
Specification intake
- Spec collector
apps/spec-collector/server.js - Collector UI
apps/spec-collector/public/app.js - Requirements volume
- Scribe agent
apps/scribe-agent/src/scribe_agent/main.py
Issue to merge
- Developer agent
apps/developer-agent/src/developer_agent/main.py - Review agent
apps/review-agent/src/review_agent/main.py
Operational response
- Fixer agent
apps/fixer-agent/src/fixer_agent/main.py - Rummager agent
apps/rummager-agent/src/rummager_agent/main.py
AI gateway
- Llama Stack gateway
apps/llama-stack-kube-github - GitHub MCP
apps/llama-stack-kube-github - Kubernetes MCP
apps/llama-stack-kube-github
Shared platform
- GitHub repository
- OpenShift cluster
- Tekton PipelineRuns
- Application logs
- vLLM inference
Other components
- Product owner
Connections 16
- Product owner → Collector UI: submits requirements
README.md - Collector UI → Spec collector: saves specification
README.md - Spec collector → Requirements volume: writes files
README.md - Requirements volume → Scribe agent: reads mounted files
README.md - Scribe agent → Llama Stack gateway: creates issues via
README.md - Developer agent → Llama Stack gateway: uses for implementation
README.md - Review agent → Llama Stack gateway: checks and merges via
README.md - Fixer agent → Llama Stack gateway: queries and opens fixes via
README.md - Rummager agent → Llama Stack gateway: investigates and files via
README.md - Llama Stack gateway → GitHub MCP: invokes tools
README.md - Llama Stack gateway → Kubernetes MCP: invokes tools
README.md - Llama Stack gateway → vLLM inference: requests inference
README.md - GitHub MCP → GitHub repository: accesses repository
README.md - Kubernetes MCP → OpenShift cluster: queries resources
README.md - OpenShift cluster → Tekton PipelineRuns: hosts pipeline runs
README.md - OpenShift cluster → Application logs: provides pod logs
README.md