jev-code
Stanley: a Jev-first coding CLI where Jev routes requests to deterministic workflows and an agent only writes new workflows when none fits.
Overview
This repository now describes Stanley, a Jev-first, self-improving command-line tool for checking and understanding code changes, previously published as the jev-code placeholder package. Instead of naming a command, you describe what you need, and Jev routes the request to one workflow: a built-in, a trusted repository workflow, or one Stanley wrote for itself. Each workflow is deterministic code that gathers bounded evidence and asks Jev small fixed-choice questions; code, not a model, makes the decisions, and this hot path never starts a general agent. Per the logicrw list, workflows cover locating code, checking change intent, triaging test failures and organising review findings.
When no workflow fits, Stanley falls back to the Pi coding agent if installed, reports the agent's own account without claiming to have verified it, and in the background asks an agent to write a new Stanley workflow for that kind of request; the candidate is validated and staged for you to promote. Version 0.1.0 is not yet on npm and must be built from source, needing Node 22.18+, git and a TypeSafe key; CI runs on Linux. It is MIT licensed.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which workflow should handle this request? | — |
| Yes / no | Does this change match its stated intent?(inferred) | yes · no |
How it works
- User describes a need in plain language
- Jev routes to a built-in, repository or self-written workflow
- Workflow gathers bounded evidence and asks Jev fixed-choice questions
- Code decides; if no workflow fits, fall back to Pi and stage a new workflow
Strengths
- Deterministic hot path with no general agent
- Self-improving workflow library with human promotion
- Honest reporting on unverified fallbacks
Caveats
- Not yet released on npm
- Windows untested
Getting started
`git clone https://github.com/devagrawal09/stanley-code.git && cd stanley-code && npm ci && npm run build && node dist/cli.js --help`.At a glance
- Type
- Project
- Status
- prototype
- Language
- TypeScript
- Runtime
- Node CLI
- Has code
- yes
- Has tests
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
Helps coding Agents locate code, check change intent, triage test failures, and organize review findings.
logicrw/awesome-jev-projects
Frequently asked questions
What does Jev decide?
Which workflow handles a request, and small fixed-choice questions within workflows.
What if no workflow fits?
It falls back to the Pi agent and stages a new workflow for you to approve.
Is it on npm?
Not yet; build from source until 0.1.0 is released.
Related projects
Gives Claude Code, Codex, Pi and OpenCode five typed Jev tools (classify, check, score, rank, ask) plus a skill on when to use them.
MCP serverbetaTypeScriptSix evidence-oriented Jev tools for pi, omp and any MCP client: ask about code, find files, locate ranges, check diffs and select tests.
ServicebetaTypeScriptJev-first CLI coding agent that routes a request to a deterministic workflow which asks Jev fixed-choice questions about bounded evidence.
ProjectprototypeTypeScriptRust code-review gate asking Jev small typed questions about functions and tests, failing CI only on rules measured 80% right.
CLIproductionRustMCP server exposing Jev Noul, Choice and Score via OpenRouter so coding agents can make quick decisions without reasoning tokens.
MCP serverbetaTypeScriptOpenAPI diff tool that adds Jev semantic review of changed prose to deterministic checks, then code decides pass, review or block.
ServicebetaPython