vibecheck-linear-jev
Linear-backed work ledger for coding agents where a Jev-compatible model verifies done reports, blockers and plan changes against the task.
Overview
vibecheck-linear-jev is a work ledger for Claude Code, Codex, Muse and Antigravity agents that keeps its record in Linear: the plan is a Linear project, each task an issue, and claims, reports and verdicts are comments and labels. Agents change task state through eleven MCP tools. A Jev-compatible judgment model then reads what they report against what was asked: a done report against the issue's goal and acceptance criteria, a blocker against the reasons the project accepts, and a plan change against the work it replaces.
Project status shows each task as reported and as verified, leading with what needs attention. The plugin also includes a skill and hooks that check the agent's own turns: briefs sent to subagents, shell deletes and the reply it stops on. It runs on Bun and can use hosted Jev or an open Jev model. It is version 0.1.0, derived from vibecheck-jev, and untested on Windows.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Yes / no | Does this done report satisfy the issue's goal and acceptance criteria? | yes · no |
| Yes / no | Is this blocker one of the reasons the project accepts? | yes · no |
| Yes / no | Is this shell delete safe / within scope?(inferred) | yes · no |
How it works
- Publish a plan to Linear as a project with issues.
- Agents claim and report work through MCP tools.
- The judgment model verifies reports, blockers and plan changes.
- Status shows reported versus verified; hooks check agent turns.
Strengths
- Separates what agents report from what is verified.
- Uses Linear as a shared, human-readable record.
- Works with hosted or open Jev models.
Caveats
- Version 0.1.0; Windows untested.
- Requires Bun and a Linear workspace.
At a glance
- Type
- Project
- Status
- prototype
- Language
- JavaScript
- Runtime
- Bun MCP server + hooks + skill
- Has code
- yes
Use cases and integrations
How curated lists describe it
A coding-agent work ledger backed by Linear: agents plan, claim and report work…
wh000wh000/awesome-jev-live
Frequently asked questions
What does the judgment model check?
Done reports against acceptance criteria, blockers against accepted reasons, and plan changes against replaced work.
Where is the record kept?
In Linear, as projects, issues, comments and labels.
Can I use an open model?
Yes, a Jev-compatible open model can be used.
Related projects
Desktop app running coding agents in parallel git worktrees, with optional Jev judgments of each changed file against the task.
AppbetaLocal desktop pal that verifies what a coding agent really did, with optional Laya or Jev double-checks of unclear test results.
ProjectbetaTypeScriptPortable skill pack giving agents a conservative Jev second opinion on intent, completion, routing and bounded browser navigation.
ProjectbetaPythonClaude Code skills that wait for AI pull-request reviewers and judge whether their reports actually permit a merge, never inferring a pass.
ProjectbetaMCP server giving coding agents fast typed decisions (ship gates, risk checks, file triage, page checks) as accept, review or escalate.
MCP serverbetaTypeScriptDeepSeek Harness plugin checking whether an agent's 'done' claim is true, using deterministic evidence plus one batched Jev assessment.
PluginbetaTypeScript