Codex Radar
GitHub Methodology
CODEX CLI PLUGIN

Score how you
collaborate with Codex.

Codex Radar reads your local Codex sessions and grades them across 9 dimensions in 3 categories — Communication, Engineering, Outcome. Your own Codex model scores the sessions and writes the diagnosis, with paste-ready improvement prompts — all in a single-file HTML dashboard.

Get started View on GitHub
✓Runs in your Codex session ✓Evidence-grounded ✓100% local
Codex Radar
sample report
FEATURE BUILD
Strong, steady delivery — but tests rarely back the changes you ship.
B
72 / 100
Comm
68B
Eng
76A
Out
73B
Strongest: Efficiency Bottleneck: Proof Check
Why Codex Radar

Grounded in your real work — not a quiz

It reads the sessions you already had with Codex, then turns them into a fair, evidence-grounded score and a concrete plan.

Reads your real sessions

Parses the actual JSONL Codex writes — every prompt, shell command and exit code, apply_patch edit, plan update, and MCP / search / subagent call.

Your Codex writes the diagnosis

A deterministic parser extracts the facts; then your own Codex model scores them against a transparent rubric — a formula baseline plus a bounded, evidence-cited adjustment — and writes a free-form coaching note.

Paste-ready prompt rewrites

Every weak signal becomes a concrete prompt you can paste into your next thread — tagged with the dimension it lifts and its expected impact.

9 Dimensions × 3 Categories

Measured, not guessed

Every score traces back to a formula and a countable signal from your session. No black box.

Communication
3 DIMS
How clearly you direct Codex through text.
Lock-On 瞄准力
Goal, expected behavior, file paths, constraints in your directives.
Scene Setting 铺场力
How much useful context your opening messages carry.
Steering 校准力
Quality of mid-session course-correction and verification asks.
Engineering
3 DIMS
How well you drive the Codex platform.
Toolcraft 工具调度
Shell, patch, plan, MCP, search, subagents + command success rate.
Architecture 工程脚手架
Durable setup — AGENTS.md, repo scaffolding, tests, manifests.
Tempo 推进节奏
Steady progress without aborts, compactions, or tool churn.
Outcome
3 DIMS
What actually gets done.
Efficiency 产出效率
Useful output per human message and per tool call.
Proof Check 验证意识
Tests, builds, lint, screenshots, renders, explicit checks.
Completion 闭环完成
Turns finish cleanly instead of trailing off mid-task.
How it works

Facts in, judgment out

Three stages, all on your machine: a deterministic parser extracts the facts, your Codex model applies the rubric, and a renderer builds the dashboard.

1

Detect & parse

Scans your local sessions, matches the current project, and extracts countable signals.

~/.codex/sessions
2

Score & diagnose

Your Codex applies the rubric's formula baseline, adds a bounded evidence-based adjustment, and writes the diagnosis.

rubric.json
3

Render

Writes a self-contained HTML dashboard you can open, scan, and share.

~/.codex-radar/reports
Fairness

A 3-message fix isn't a 50-session build

Every project is auto-classified into one of four profiles, each with its own category weights.

AOne-shot · few messages, light editing
BFeature build · real implementation work
CLong-running · many sessions or compactions
DLearning · conversation-heavy, few edits
Honest N/A, not a faked 50. When a signal genuinely can't be evaluated — like Architecture when the project directory no longer exists on this machine — the dimension is marked N/A and dropped from the average.
Privacy

100% on your machine

Your session data never leaves your computer. No network calls, ever.

No cloud
No separate API key
Runs in your Codex
No telemetry
Read-only sessions
Get started

Two commands, then ask

Requires the Codex CLI and Node.js 18+. No build step, no server.

codex
$ codex plugin marketplace add LeifDiao/codex-radar
$ codex plugin add codex-radar@codex-radar-marketplace

# start a new thread, then ask:
› Run Codex Radar on this project

Or develop locally: git clone the repo, then codex plugin marketplace add .

See how you really work with Codex

A few seconds of pure local analysis. No tokens, no cloud, no waiting on a model.

Get Codex Radar