Week 27 · 29 Jun–5 Jul 2026
A first proof of concept connected an agent to the workflow engine, with its contract and a specification for tool authorization.
· Written
- PRs opened
- 4
- PRs merged
- 3
- PRs reviewed
- 0
- Issues resolved
- 0
- Active repositories
- 3 of 5
Counts cover 29 Jun–5 Jul 2026 (UTC) across the project's tracked repositories. Pull request figures count only pull requests opened during these dates; merged, closed and reviewed mean that also happened within them, so one can count in more than one category. They count pull requests of any size, including many prepared by AI coding agents and some fully automated ones. Reviewed means a submitted review by someone other than the author, not approval.
This week we built the first proof of concept that connects an agent to the workflow engine. The agent runs on LangGraph and reaches the workflow engine through a governed bridge, under a workflow contract we wrote for it, and we specified how agents are authorised to use tools. The ninth and tenth posts in the blog series were also produced. Written retrospectively from the project's history.
What we worked on
- Agents and runtime. We built a proof-of-concept agent on LangGraph, connected to the workflow engine through a governed bridge, and improved its security and reliability.
- Platform specification. We wrote the proof of concept's workflow contract and specified how agents are authorised to use tools.
- Domain packs. We drafted a database schema for trading signals and orders for the proof of concept.
- Content. We produced the ninth and tenth posts in the blog series, on the twelve-layer architecture and on memory.
29-Jun-2026
- Agents and runtime. We built a proof-of-concept agent on LangGraph, connected to the workflow engine through a governed bridge.
- Platform specification. We wrote the proof of concept's workflow contract and updated the technology stack to match.
- Content. We produced the ninth post in the blog series, on the twelve-layer architecture.
01-Jul-2026
- Platform specification. We specified how agents are authorised to use tools.
02-Jul-2026
- Agents and runtime. The proof-of-concept agent's security, performance and reliability were improved.
- Domain packs. We drafted a database schema for trading signals and orders for the proof of concept.
03-Jul-2026
- Content. We prepared the tenth post in the blog series, on memory.
Still in progress
- The proof of concept is being prepared for an end-to-end run.
- The next posts in the blog series are being prepared.
Looking ahead
Next we turn to running the proof of concept end to end. The tool authorization specification continues to be reviewed.