Files
lda-wf/readme.md
T

2.4 KiB

lda.chat workflow platform

This repository is building a workflow runtime for AI-assisted digital work. Agents discover capabilities, author typed workflows, validate them, deploy them, and run them through a deterministic executor. The LLM plans; the runtime executes.

Quick Start

uv run wf --help
uv run wf --local cap list --format ids
uv run wf --local cap inspect wf.std.constant
uv run wf --local cap call wf.std.constant --input '{"value":"hello"}'
uv run pytest -q

Run the durable JSON-RPC server:

uv run wf-rpc-server --config wf.config.json --host 127.0.0.1 --port 8765
uv run wf --url http://127.0.0.1:8765/rpc cap list --format ids

Current Shape

wf_cli
  -> wf_transport_rpc_http or local target
  -> wf_server
  -> wf_api
  -> wf_core / wf_artifacts / wf_sources_*
  • wf_core: deterministic graph/runtime kernel.
  • wf_authoring: Python authoring helpers and NodeSpec creation.
  • wf_api: workflow application surface shared by local and remote frontends.
  • wf_server: durable server composition boundary.
  • wf_transport_rpc_http: JSON-RPC-over-HTTP client/server transport.
  • wf_sources_mcp: upstream MCP source implementation and persistent runtime.
  • wf_mcp: MCP frontend, legacy entrypoints, proxy/admin glue, and compatibility shims during extraction.
  • wf_cli: command-line frontend over local or remote workflow APIs.

Start Reading

Historical design notes and completed implementation plans live under docs/historical/. The previous long top-level running-design note is intentionally no longer the project front door; current model details are split across the docs index above.