A change is committed where it is made
Agent workflow rules now say when to commit and push. A new change is committed and pushed as it is made. A card of a change's own worktree offers Commit Change, and the git stage commits what the stages left.
Agent workflow rules now say when to commit and push. A new change is committed and pushed as it is made. A card of a change's own worktree offers Commit Change, and the git stage commits what the stages left.
What the product says to a person is now written in one register, every message carries an OSW-XXX code, and a generated page lets you look up the identifier you saw.
An agent can now raise a question for the operator, the chain stops and waits, answers are kept in decisions.md, and the product's own agent asks in its turn from the start.
Two new shared core tools, search_web and fetch_webpage, let local-model agents consult a SearXNG endpoint and retrieve HTML pages as Markdown. No remote agents are changed.
OpenSpec Workbench now ships a versioned catalogue of rules, checks that run only when the host is idle, and findings that explain what is wrong without moving a single file.
A chain has four stages you can hand to four different agents, at four different prices. The harness can prove the savings on two of the three agents involved, and is honestly blind to the third.
A 54 MB leftover that survived a whole day of cleanup passes, and a context window nobody was reading until DeepSeek's own agent told us how full it was.
Adding DeepSeek as a tenth agent touched one file in the core, and the finding worth reading is what dsh does on the wrong Node.
OpenSpec Workbench now has an Owner, an Implementer, a signed history, and a board with six stages, all read from git and the forge, with no server and no database.
GitHub, GitLab or Gitea now sit behind one Forge interface. What that took, and the quirks live testing found on each of the three.
How OpenSpec Workbench keeps a VS Code extension and a local web application from drifting apart, what was rejected on the way, and where the "thin adapter" idea turned out not to be free.
OpenSpec Workbench starts a coding agent on a change, shows what it is doing while it works, and lets you stop it where its work is sound. A tour of the Pipeline, the agents it drives and the limits it keeps.
Showing an OpenSpec change is a solved problem. Supervising the agent that builds it is not. Three things that went wrong while building OpenSpec Workbench, and what a cockpit has to do about them.