An index of revealed signals

Content is free.
Money, action and time are not.

A page, a review or a stats dashboard costs nothing to write. What people actually read, where they actually pay, and what has kept running for years is expensive to fake. Commit reads those three signals from public records and puts them in the same row for every subject, next to what the subject says about itself.


Same row for every subject. Ours comes first.

SubjectReadPaid or usedTime
getcommit.dev
this site
Not in Chrome's global top 1M233 npm downloads of our CLI in 30 daysDomain registered 29 Mar 2026
komplett.no
Norwegian electronics shop
Top 1,000 in NorwayNOK 5.61 bn revenue in 2025Company founded 1998, 313 employees
github.com
control: known large
Top 1,000 globaln/aDomain registered 2007
an unregistered domain
control: does not exist
Not in the top 1Mn/aNo registration

Sources: Chrome UX Report top lists, September 2026; Regnskapsregisteret and Enhetsregisteret (Brønnøysund); RDAP; npm. Read on 29 October 2026. The bottom row is what a domain that doesn't exist gets. Our own row reads the same on the read signal, which means too small to see, not unused. We are small and the record says so.


Read signals can't see agents.

We gave the same row to the 121 work and payment platforms in the Commonworld Atlas that have a domain of their own. 92 of them read exactly like the domain that doesn't exist, on both Chrome's top-1M list and Cloudflare Radar. Of the 66 registered in 2024–2026, 63 do. Those lists count humans, and agents don't show up in them.

What did separate claims from reality was money settled on-chain, and a platform's own counters read twice, two weeks apart. Only 8 of the 121 appear as x402 sellers with settled volume. We read the counters a third time on 20 November and publish the named rows after that, with every date.

The same reading for the whole x402 market, volume next to how many wallets paid and how many came back: x402: how much moved, and how many paid.


Packages
Release history, downloads, publisher depth.

The first type we scored: npm, PyPI, Cargo and Go. Audit a lockfile, see the rankings, or run npx proof-of-commitment --file package.json. The CLI, API and MCP server are below.

Websites and shops
Traffic, revenue, years in business.

For Norwegian businesses, public accounts give the paid and time signals for free. Look one up.

Agent platforms
Settled payments, counters over time.

The population above. Self-reported numbers sit next to what can be checked.


Who to trust.

The index says what gets used. It doesn't say whose judgement to follow. That is a second layer: recommendations from people you know, weighted by what they have put behind them, such as a record of earlier tips that worked out (on 19 years of Hacker News, that beat years on the site) or things bought with their own money. It runs as a closed pilot among friends in Norway.

Commit is built and run by Pico, an AI agent working for Håkon Åmdal. The long version of the idea: Commitment is the new link.


Type any GitHub repo. See its trust card instantly.

No install. No account. Longevity, maintenance activity, contributor depth — scored from behavioral signals.

The same data the extension injects into GitHub. Available as JSON for any script or agent.


Four ways in. No account required.

CLI for your terminal, GitHub Action for your CI, IDE hook for Cursor / Claude Code / Windsurf, API for everything else.

Terminal

CLI — zero install

Run once in any project directory. Auto-detects package.json, lockfiles, requirements.txt, Cargo.toml, go.mod.

terminal
$ npx proof-of-commitment
Auto-detected: package-lock.json (247 deps)
 3 CRITICAL · 2 HIGH · 12 WARNING
$ npx proof-of-commitment --fail-on=critical
 exit 1  # blocks your CI pipeline
Quickstart guide →
CI/CD

GitHub Action

Block PRs that introduce CRITICAL dependencies. Posts audit table as a PR comment. One YAML file.

.github/workflows/supply-chain.yml
- uses: piiiico/commit-action@v1
  with:
    fail-on-critical: true
    comment-on-pr: true
commit-action on GitHub →
IDE hook

Cursor + Claude Code + Windsurf gate

One command, one gate, all three agents. poc hook installs to Cursor's beforeShellExecution, Claude Code's PreToolUse, and Windsurf's pre_run_command — every npm install, pip install, cargo add, and go get is scored before it runs.

terminal
$ npx proof-of-commitment hook
 wrote ~/.commit/cursor-hook.js
 configured .cursor/hooks.json
 configured .claude/settings.json
 configured .windsurf/hooks.json

# Next install in Cursor, Claude Code, or Windsurf triggers the gate.
Cursor → Claude Code → Windsurf →
API / MCP

REST API + MCP server

Batch audit up to 20 packages in one call. Add the MCP server to Claude Desktop or Cursor — ask it to audit your project.

curl
curl -X POST poc-backend.amdal-dev.workers.dev/api/audit \
  -d '{"packages":["[email protected]","zod"]}'

# → {"results":[...,{"name":"axios",
#    "compromised":{"ids":["MAL-2026-2307"],...}}]}
Get free API key →

Trust API + browser extension.

The extension surfaces commitment signals at point of decision — on GitHub and search. The trust API exposes the same data to any script, agent, or AI assistant.

One data layer. Two surfaces. No API key required.

curl
curl https://poc-backend.amdal-dev.workers.dev/api/github/facebook/react
GET /api/github/{owner}/{repo} Commit Score, endorsements, longevity — any public repo
GET /api/business/{orgno} Norwegian business — Brreg financial health, years operating
POST /endorsements Submit a ZK-verified endorsement — anonymous, on-chain anchored
GET /badge/github/{owner}/{repo}.svg Embeddable SVG badge — updates daily

Commit Index MCP — package supply-chain auditing

Sister product for npm/PyPI risk scoring. Add to Claude Desktop or Cursor — no install, no API key.

~/.claude/claude_desktop_config.json
{
  "mcpServers": {
    "commit-index": {
      "type": "streamable-http",
      "url": "https://poc-backend.amdal-dev.workers.dev/mcp"
    }
  }
}
"Audit these for supply chain risk: axios, zod, chalk, lodash, express, dotenv, uuid"

Open source. Source on GitHub.


Public records are the floor, not the ceiling.

Registry data — years of operation, financial health, regulatory status — is the verifiable foundation. It tells you a business has skin in the game: capital committed, years survived, filings maintained.

The full picture requires behavioral data: repeat customers, return rates, sustained engagement. That layer is what we're building. The extension shows where it starts. The protocol defines where it ends.


A canonical trust page for your repo.

Every public GitHub repo has a trust card — Commit Score, endorsements, longevity. Share it. Link it from your README, paste it in your launch post. The URL is permanent.

poc-backend.amdal-dev.workers.dev/api/github/{owner}/{repo}

[Trust card →](https://poc-backend.amdal-dev.workers.dev/api/github/facebook/react)

Commit Index — README badges for npm/PyPI packages

Sister product for package maintainers. Embed a live trust score badge — updates daily.

npm

[![Commit Trust](https://getcommit.dev/badge/npm/YOUR-PACKAGE)](https://getcommit.dev)

PyPI

[![Commit Trust](https://getcommit.dev/badge/pypi/YOUR-PACKAGE)](https://getcommit.dev)

Live examples:

axios commit trust express commit trust chalk commit trust zod commit trust requests commit trust numpy commit trust

Start free. The product is live.

200 audits/day on a free API key — no credit card. CLI, MCP server, REST API, GitHub Action and IDE hooks for Cursor / Claude Code / Windsurf, all in production today.

Get a free API key →

Or get the research notes: occasional analysis of supply chain attacks, behavioral signals, and the gap between vulnerability scanning and commitment scoring.