# oh-my-knowledge

[![npm version](https://img.shields.io/npm/v/oh-my-knowledge.svg)](https://www.npmjs.com/package/oh-my-knowledge)
[![npm weekly downloads](https://img.shields.io/npm/dw/oh-my-knowledge.svg)](https://www.npmjs.com/package/oh-my-knowledge)
[![CI](https://github.com/lizhiyao/oh-my-knowledge/actions/workflows/ci.yml/badge.svg)](https://github.com/lizhiyao/oh-my-knowledge/actions/workflows/ci.yml)
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](./LICENSE)
[![Node.js Version](https://img.shields.io/node/v/oh-my-knowledge.svg)](https://nodejs.org)

**English** | [简体中文](./README.zh.md)

**Stop editing LLM knowledge inputs by gut feel.**
`oh-my-knowledge` (omk) is a measurement workflow for prompts, RAG context, skills, agents, and workflows. It fixes the executor model and the evaluation samples, changes only the knowledge artifact, then answers the release question that matters: **can v2 ship, and where is it better?**

![omk knowledge artifact evaluation flow: doctor / eval / observe / sample / evolve loop](./docs/public/omk-knowledge-flow-en-animated.gif)

📖 **Full documentation: [oh-my-knowledge.pages.dev](https://oh-my-knowledge.pages.dev)** (searchable, English / 简体中文)

## What omk makes measurable

| Decision | Command | Evidence you get |
|---|---|---|
| Is this artifact coherent enough to evaluate? | `omk doctor` | structure, dependencies, safety, and measurability checks |
| Is v2 actually better than v1? | `omk eval` | one-line verdict, confidence interval, failed samples, cost |
| Why did it pass or fail? | `omk studio` | report view with scores, diagnostics, and examples |
| Should this version become the accepted one? | `omk promote` / `omk evolve` | evidence-gated accept or generate a better candidate |
| What did real usage expose? | `omk observe` / `omk sample --from-traces` | production gaps drafted for review; reviewed drafts can become eval samples |

![omk report — verdict pill "v2 is clearly better than v1 — ready to ship"](./assets/screenshots/report-overview.png)

## Quick start

```bash
npm i -g oh-my-knowledge
omk init demo && cd demo
omk eval --control code-review-v1 --treatment code-review-v2 --dry-run
omk eval --control code-review-v1 --treatment code-review-v2
```

Runs out of the box — no edits needed first. `omk init` scaffolds two skill variants and three sample cases; `--dry-run` previews calls and cost; `omk eval` runs the controlled A/B and opens an HTML report with a one-line verdict in about five minutes. Once it runs, swap in your own skills and cases.

Prerequisite: configure one authenticated model runtime (Codex CLI, Claude Code, or an API executor; see [Requirements](#requirements)). Inside a Codex task in the ChatGPT desktop app, omk automatically selects `codex`, reads the model from `~/.codex/config.toml`, and uses the same Codex model as the default judge. Claude is not required.

To make Codex the default in regular terminals, add the preference to your shell profile (for example `~/.zshrc`):

```bash
export OMK_EXECUTOR=codex
# Optional: export OMK_MODEL="your-codex-model"
```

Without `OMK_MODEL`, omk reads the model from `~/.codex/config.toml`. You can still pass `--executor codex --model <codex-model>` per command. Pass `--judge-models` or set `OMK_JUDGE_MODELS` only when you want a different judge.

> The first run has only 3 cases, so the verdict will usually be `UNDERPOWERED` (insufficient data) — that's a normal starting point, not an error; grow to ~20+ cases before trusting a ship/no-ship call.

> The CLI notifies you when a newer version is available (at most once per 20h); set `OMK_SKIP_UPDATE_CHECK=1` to silence it permanently.

Walkthrough: [5-minute quickstart guide](docs/quickstart-skill-eval.md) (recommended for first-time users; includes demo → own skill → verdict actions). More runnable examples (Skill Map, A/B, offline executor, agent runtime, RAG) live in the repo's [example gallery](https://github.com/lizhiyao/oh-my-knowledge/tree/main/examples).

Deeper: [who omk is for](docs/explanation/who-omk-is-for.md) · [CLI reference](docs/reference/cli.md) · [how it works](docs/explanation/architecture.md) · [eval sample format](docs/reference/eval-sample-format.md) · [executors](docs/reference/executors.md) · [artifact layout](docs/reference/artifact-layout.md)

## The omk loop

omk is for authors and maintainers of LLM knowledge artifacts who need a release decision, not for passive end-users of a skill. The main loop is deliberately controlled:

```text
change a prompt / RAG / skill / agent artifact
→ run omk doctor before evaluation
→ run omk eval with the same model and the same samples
→ read the report / Studio evidence
→ promote a proven version or evolve a candidate
→ observe real usage and draft gap-derived samples for review
```

The first value is the pre-ship `doctor → eval` decision. The long-term value is the closed loop: `observe` surfaces production gaps, `sample --from-traces` drafts regression samples for human review, and reviewed drafts can become fixed eval samples that make the next `eval` harder to game.

## Use inside AI Coding Agents

Install the official omk Agent Skill to let your coding agent run omk workflows from natural language:

```bash
omk install omk-agent-skill
```

By default, omk installs only into detected local targets it explicitly supports: Codex/AGENTS when `~/.codex` or `~/.agents` exists, and Claude Code when `~/.claude` exists. Use `--to all` to force every target omk currently knows, or `--dest` for a custom skill root.

### Use inside Claude Code

When the `omk` skill is available in Claude Code, you can invoke it directly:

```bash
/omk eval              # evaluate the artifact(s) in the current project
/omk evolve            # auto-iterate to improve a skill
/omk sample            # generate or fill test cases
```

These slash commands are natural-language entry points — the agent reads the conversation context to figure out which skill to operate on. You can also just say "compare v1 vs v2 for me" or "improve this artifact" and omk picks the right command.

### Use inside Codex

Codex does not support Claude Code style `/omk ...` slash commands. Ask the agent to run the `omk` CLI directly. Inside a Codex task, omk automatically selects the Codex runtime and locally configured model:

```bash
omk eval
omk evolve skills/my-skill.md   # one-shot: doctor → (auto-generate samples if missing) → self-iterate
omk sample skills/my-skill.md
```

You can also describe the goal in natural language, such as "compare v1 vs v2" or "generate test cases for this skill".

`eval`, `doctor`, `sample`, `evolve`, and the LLM-enhanced observe review share the same runtime resolution. Once Codex is selected, the default judge reuses the evaluated Codex model instead of falling back to `claude:haiku`.

> `omk evolve` is a one-shot loop: it runs the doctor gate first, auto-generates eval samples when the target skill has none, then self-iterates. For a brand-new skill, just run `omk evolve skills/foo.md`.

## Why this tool

Knowledge engineering creates a versioning problem: every prompt, RAG recipe, skill, agent, or workflow can change behavior without changing application code. When someone asks "can we ship v2, and why?", a prettier answer or a higher anecdotal success rate is not enough.

omk treats the knowledge artifact as the variable under test: **same model, same evaluation samples, only the artifact changes.** That makes the comparison explainable, repeatable, and suitable for CI or release review.

## Why omk over alternatives

| | omk | promptfoo | DeepEval | LangSmith |
|--|--|--|--|--|
| Bootstrap CI | ✓ default | ✗ | ✗ | ✗ |
| Krippendorff α (judge ↔ human) | ✓ with gold set | ✗ | ✗ | ✗ |
| Length-debias judge prompt | ✓ default | ✗ | ✗ | ✗ |
| Saturation curve | ✓ | ✗ | ✗ | ✗ |
| Three-layer scoring isolation | ✓ | ✗ | partial | ✗ |
| Per-variant skill isolation (construct validity) | ✓ default | ✗ | ✗ | ✗ |
| Native Agent Skill | ✓ | ✗ | ✗ | ✗ |
| Hosted SaaS dashboard | ✗ | ✗ | ✓ | ✓ |

omk's moat is **default-on safety net** — Bootstrap CI and length-debias aren't advanced flags; they're the default, and judge ↔ human α comes free the moment you add a gold set. Other tools let you opt into confidence intervals; omk makes them unavoidable. Need a hosted SaaS dashboard? Choose LangSmith. Want quick local prompt iteration without statistics? Choose promptfoo. **Shipping to production and someone will ask "why should I trust this number?" Choose omk.**

RAG-specific evals: see RAGAS (separate niche, complementary to omk). Full comparison with 7 tools across 25+ dimensions: [docs/reference/comparison.md](docs/reference/comparison.md).

## Features

| Feature | What it does |
|---|---|
| **One-line verdict** | `omk eval` six-tier verdict + ship recommendation + exit-code routing; HTML pill shares the same rules |
| **Six-dim evaluation** | Fact / Behavior / LLM-judge / Cost / Efficiency / Stability shown independently |
| **Multi-executor** | Claude CLI / Claude SDK / Codex CLI / Codex SDK / OpenAI / Gemini / Anthropic API / any custom command |
| **30+ assertion types** | substring, regex, JSON Schema, ROUGE/BLEU/Levenshtein similarity, agent tool-call assertions, semantic similarity, custom JS |
| **Statistical rigor** | Bootstrap CI / length-debias / saturation curve on by default; Krippendorff α auto-computed with a gold set. [Details →](docs/explanation/statistical-rigor.md) |
| **RAG metrics** | `faithfulness` / `answer_relevancy` / `context_recall` — anti-hallucination + answer relevance + context coverage |
| **LLM health audit** | `omk doctor` grades 7 builtin dimensions; repeats the audit (`--repeat`) and merges findings by k/n consensus |
| **Production observability** | normalize Codex, Claude Code, OpenClaw, and markdown logs into source-neutral Trace IR; measure per-skill outcomes / latency / token use / knowledge-gap signals |
| **Knowledge-gap detection** | severity-weighted signals quantify risk exposure instead of claiming completeness |
| **Construct-validity isolation** | `--strict-baseline` (default ON) cuts three contamination channels so baseline doesn't silently see the skill it's being compared against |
| **Git & remote sources** | install / eval from a local git ref or a remote git URL (`--git-url`); directory-skills run in a content-addressed **isolated copy** so `references/` assets are real measured input, not just `SKILL.md` |
| **Evidence-gated management** | `omk install` registers a managed record; `omk eval` auto-writes evidence bound by content fingerprint, moving a skill `installed → measurable`; `omk list` surfaces each managed skill's status (installed / measurable / promoted / stale); `omk promote` accepts a version once its evidence passes the gate (default PROGRESS only); `omk rollback` revokes that acceptance, returning the skill to `measurable`. [spec →](docs/specs/evidence-gated-management.md) |
| **Sample design science** | sample schema with `capability` / `difficulty` / `construct` / `provenance` metadata (HF Dataset Cards style); studio surfaces coverage breakdown plus `rubric_clarity_low` / `capability_thin` flags. [docs/specs/sample-design-spec.md](docs/specs/sample-design-spec.md) |
| **Multi-judge ensemble** | `--judge-models claude:opus,openai-api:gpt-4o` cross-vendor scoring + agreement metrics |
| **Multi-run variance** | `--repeat N` repeats the eval and computes mean / SD / CI / t-test |
| **MCP URL fetching** | pull content from private-doc URLs via an MCP server (SSO-protected knowledge bases, etc.) |
| **Auto analysis** | detects low-discrimination assertions, flat scores, all-pass / all-fail, expensive samples |
| **Traceability** | reports carry CLI version, Node version, artifact version fingerprint, judge prompt hash |
| **EN / ZH switch** | one-click language toggle in the HTML report |

## Documentation

The full docs are published at **[oh-my-knowledge.pages.dev](https://oh-my-knowledge.pages.dev)** — searchable, with an English / 简体中文 switcher. Key pages:

- **[How it works](docs/explanation/architecture.md)** — interleaved scheduling, variant resolution, dual-channel scoring, six-dim report
- **[Eval sample format](docs/reference/eval-sample-format.md)** — sample schema, scoring formulas, 30+ assertion types, custom JS assertions
- **[CLI reference](docs/reference/cli.md)** — all top-level commands with bash examples and flag tables
- **[Executors](docs/reference/executors.md)** & **[artifact layout](docs/reference/artifact-layout.md)** — built-in / custom executors; how `variant` resolves to an artifact + runtime context
- **[How-to guides](docs/guides/agent-eval.md)** — [evaluate an agent](docs/guides/agent-eval.md) (project runtime context) and [use non-Claude models](docs/guides/non-claude-models.md) (GLM / Qwen / DeepSeek / Moonshot / Ollama)
- **[Quickstart](docs/quickstart-skill-eval.md)** — first-time five-minute walkthrough
- **[Example gallery](https://github.com/lizhiyao/oh-my-knowledge/tree/main/examples)** — a set of runnable examples in the repo, arranged simplest-to-richest
- **[Sample design spec](docs/specs/sample-design-spec.md)** — capability / construct / provenance metadata; industry-gap mapping
- **[Statistical rigor](docs/explanation/statistical-rigor.md)** — why bootstrap CI / α / length-debias / saturation matter
- **[Comparison with 7 tools](docs/reference/comparison.md)** — 25+ dimensions across promptfoo / DeepEval / RAGAS / OpenAI Evals / LangSmith / lm-eval-harness / inspect-ai
- **[Evidence-gated management](docs/specs/evidence-gated-management.md)** — managed records, lifecycle states (installed / measurable / promoted / stale), install → eval → measurable → promote → rollback

## Environment variables

| Variable | Description |
|---|---|
| `OMK_EXECUTOR` | default executor preference, e.g. `codex` / `codex-sdk` / `claude` |
| `OMK_MODEL` | default evaluated model; Codex reads local `config.toml` when unset |
| `OMK_JUDGE_MODELS` | default judge list in `executor:model[,...]` format |
| `CCV_PROXY_URL` | proxy requests through cc-viewer for live eval-traffic visualization |
| `OMK_REPORT_PORT` | report server port (default: 7799) |

## Requirements

- Node.js >= 22
- At least one authenticated model runtime:
  - Codex: install and authenticate the Codex CLI (`npm i -g @openai/codex`); Codex tasks in the ChatGPT desktop app select it automatically
  - Claude: install and authenticate [Claude Code](https://claude.ai/code)
  - API / other executors: configure them as described in [Executors](docs/reference/executors.md)

## Security notice

This tool is designed for **local trusted environments** (dev machines, CI pipelines). The following features execute local code — make sure inputs come from a trusted source:

| Feature | Risk | Scope |
|---|---|---|
| **Custom assertions** (`custom`) | dynamically loads and executes user-specified `.mjs` files | only use assertion files you authored or reviewed |
| **eval-samples.json** | assertion configs can reference external file paths | don't use sample files from untrusted sources |

**Recommendations:**

- Do not expose the local report server on the public internet (no auth)
- Don't use third-party eval-samples you haven't vetted
- Custom assertions have a 30-second timeout but no sandbox isolation

---

See [GitHub Releases](https://github.com/lizhiyao/oh-my-knowledge/releases) for release notes. Contributions welcome — see [CONTRIBUTING](./CONTRIBUTING.md).