OPENSPEC V1.10 PLUGS INTO ZED’S AGENT AND FORCES REAL “DONE” CRITERIA
OpenSpec v1.10.0 now plugs into Zed’s Agent and forces task plans to declare a real definition of done. The [v1.10.0 release](https://github.com/Fission-AI/Ope...
OpenSpec v1.10.0 now plugs into Zed’s Agent and forces task plans to declare a real definition of done.
The v1.10.0 release adds a Zed Agent integration via openspec init --tools zed (Zed ≥1.4.2), a quieter npm global install, and a change that requires each generated task plan to state how it’s considered done (test, command, observable result, or delivered artifact).
That focus on observable completion lines up with hard-won ops lessons from this write-up on launchd pitfalls at automation scale, where silent restarts and scattered logs masked dead jobs until an environment map and daily git diffs exposed drift.
Clear, testable 'done' criteria reduce ambiguous tasks and failed handoffs in specs and tickets.
Zed Agent support brings spec-driven suggestions directly into the editor workflow.
-
terminal
In Zed ≥1.4.2, run
openspec init --tools zedand verify/openspec-proposeappears and executes against a repo. -
terminal
Generate a task plan and confirm the 'definition of done' fields are concrete enough to wire into CI or health checks.
Legacy codebase integration strategies...
- 01.
Adopt OpenSpec’s 'done' fields to tighten existing runbooks and tickets, then map them to current CI jobs or monitors.
- 02.
Create a lightweight job inventory page (like the environment map) and commit it daily to catch drift via git diffs.
Fresh architecture paradigms...
- 01.
Start with spec-first tasks that declare tests or observable outcomes, then scaffold implementation from the editor via Zed Agent.
- 02.
Design pipelines so each step exposes a measurable signal that matches the spec’s 'done' clause.
Get daily OPENSPEC + SDLC updates.
- Practical tactics you can ship tomorrow
- Tooling, workflows, and architecture notes
- One short email each weekday