Skip to content

docs: README, architecture docs, contributing guide, and CI#4

Merged
adel-saoud merged 1 commit into
mainfrom
feat/docs
May 23, 2026
Merged

docs: README, architecture docs, contributing guide, and CI#4
adel-saoud merged 1 commit into
mainfrom
feat/docs

Conversation

@adel-saoud

Copy link
Copy Markdown
Owner

Summary

Adds the full documentation layer and GitHub Actions CI pipeline.

README highlights:

  • Feature table surfacing all 8 capabilities upfront (RAG attribution, circuit breaker, fallback routing, rate limiting, budget caps, admin API, dashboards, OpenAI compatibility)
  • demo.gif — terminal recording of the full RAG pipeline + load test cost savings output
  • Three live Grafana dashboard screenshots populated by make demo-full
  • Response header reference (X-Gateway-Cost-USD, X-Gateway-Model-Used, X-Gateway-Rag-Operation)
  • Mermaid architecture diagram showing the full middleware → router → provider → ledger flow
  • "Why not LiteLLM?" comparison table covering the five design decisions that differ
  • Resilience section: circuit breaker state diagram and two-candidate fallback chain explanation
  • Honest limitations section (single-process, Ollama-only, simulated costs, SQLite single-writer)

docs/architecture.md: full module map, data-flow diagram, middleware ordering rationale, and design decision notes for pure-ASGI streaming, per-instance Prometheus registry, and SQLite-vs-Redis split.

CONTRIBUTING.md: step-by-step guides for adding a new provider adapter, a new RAG operation type, a new metric, and a storage migration.

.github/workflows/ci.yml: four gates on every push and PR — ruff lint, ruff format check, pyright strict, pytest with coverage floor.

.env.example: all GATEWAY_* environment variables documented with defaults.

Test plan

  • CI badge is green after merge
  • make check passes locally (all four gates)
  • README renders correctly on GitHub (Mermaid diagrams, images, tables)

README:
- Feature overview table (RAG attribution, circuit breaker, fallback
  routing, rate limiting, budget caps, admin API, dashboards)
- Quick-start for both bundled and host Ollama
- demo.gif terminal recording of the full RAG pipeline + load test
- Three Grafana dashboard screenshots (rag-cost, operations, teams-budget)
- Request/response header reference table
- How-it-works architecture diagram (Mermaid)
- Why-not-LiteLLM comparison table
- RAG classifier signal priority table
- Resilience section: circuit breaker state diagram and fallback chain
- Honest limitations section

docs/architecture.md:
  Full module map with responsibility descriptions, data-flow diagram,
  middleware ordering rationale, and design decision notes (pure-ASGI
  streaming, per-instance Prometheus registry, SQLite vs Redis split).

docs/demo.tape: vhs tape for regenerating demo.gif

context/: PRODUCT / PLATFORM / PROCESS project context documents

CONTRIBUTING.md: how to add a provider, a RAG operation type, a
  metric, or a storage migration.

.github/workflows/ci.yml: lint + format + type-check + pytest on
  every push and PR (ubuntu-latest, Python 3.11).

.pre-commit-config.yaml: ruff check + ruff format hooks.

.env.example: all GATEWAY_* environment variables with defaults.
@adel-saoud adel-saoud self-assigned this May 23, 2026
@adel-saoud adel-saoud merged commit fa5eea7 into main May 23, 2026
1 check failed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant