Experimental foundation for AI-native workflow primitives, observability, and agent-first dev tooling.
TTA.dev exists to build the reusable AI devops tools that the future TTA product will rely on. The repository has a real, tested core today, but it is still aspirational overall and should not yet be read as a finished product.
The commands below were re-verified in a clean clone on 2026-03-22.
# 1. Clone and install dependencies
git clone https://github.com/theinterneti/TTA.dev.git
cd TTA.dev
./setup.sh
# 2. Terminal 1: start the observability server
uv run python -m ttadev.observability
# 3. Terminal 2: generate trace data
uv run python scripts/test_realtime_traces.py
# 4. Optional: inspect the emitted spans directly
curl http://localhost:8000/api/v2/health
curl http://localhost:8000/api/v2/spans | headThat flow currently proves:
- the observability server starts on
http://localhost:8000 - a workflow can execute successfully with current primitives
- trace data is written to
.observability/traces.jsonl - the server ingests those traces and exposes them via
/api/v2/spans
- Composable primitives for sequencing, parallelism, retries, timeouts, caching, and more
- A local observability server with a working v2 API and dashboard entrypoint
- Agent and workflow foundations in
ttadev/agents/andttadev/workflows/ - An L0 developer control plane for local task, run, gate, lock, and ownership coordination
- A documented L0-backed workflow proof path via
tta workflow run feature_dev --track-l0 - A substantial automated test suite covering a large portion of the current core
- broader example coverage beyond the narrow proof path is still catching up
- some package-local and historical docs still need follow-up cleanup for full consistency
- type-checking still fails in several areas
- some integrations remain partial or stubbed
- the first documented multi-agent proof path exists, but the repo still needs broader workflow coverage and stronger validation of what is stable vs experimental
- Getting Started Guide - current, step-by-step setup and proof path
- Quickstart - shortest honest verification flow
- 📖 Observability Quickstart → — start the dashboard, watch live traces, connect Langfuse
- User Journey - vision plus current-reality framing
- Roadmap - what is implemented, partial, and still aspirational
- Primitives Catalog - available primitives and API surface
- Agent Instructions - how AI agents are expected to work in this repo
- Contributing - development guide
- Feature Dev + L0 Example - run
the current proof workflow and inspect it through
tta control - Resilient LLM pipeline — Groq wrapped in retry + circuit breaker + OTel tracing
MIT