polygraph.so

The independent trust layer for AI tools

Independent trust for the AI tools your ecosystem runs on.

Agents plug into third-party MCP servers and skills that can hijack them or leak data. polygraph.so is an independent lab that grades those tools behaviorally, publishes evidence anyone can re-run, and keeps the grade current, because a tool surface doesn’t hold still. Free to read as a public index; continuous and monitored for your network:

For ecosystems & networks

A continuous trust index

We keep an independent, reproducible index of your network’s MCP servers, agents, and skills, re-graded on a cadence, with drops and swapped tool surfaces flagged to your team.

Start monitoring See a live index

Free & public

The public trust index

Every grade we publish is free to read and reproducible. Browse behavioral grades for MCP servers and skills across networks, and see exactly what an index looks like before you run one for yours.

Browse the index Methodology

Building the tools yourself? The open litmus harness (CLI, GitHub Action, and README badges) is self-serve on the builders page →

Monitored ecosystems

continuously re-graded · become one

01/What we watch

Live indexes, read from real evidence.

Each index below is read live from the same reproducible grades the per-network pages show.

225 live grades · 6 networks

Bankr ecosystem

monitored

Static safety grades for the Bankr skill library, plus behavioral grades for the agents that ship their own MCP server.

106A1D2F
107 skills · 7 agentsLast graded 2026-08-17View ecosystem →

Base network

MCP servers an onchain agent on Base can use: the projects integrating Base MCP, plus the wider DeFi, data, and infrastructure servers around the network.

16A10B2D
28 graded · 33 trackedLast graded 2026-07-06View ecosystem →

Virtuals Protocol

Behavioral grades for the MCP surface of the Virtuals agent launchpad on Base: the protocol's own commerce and framework infrastructure, and the agents launched on it.

1A1B1D
3 graded · 21 trackedLast graded 2026-07-06View ecosystem →

Your network

Not listed yet?

Start a trust index for your ecosystem’s servers, agents, and skills: graded, then re-graded on a cadence.

Start monitoring

See all 6 ecosystems →

02/Why it holds

Continuous, and impossible to buy.

Continuous

A grade is a measurement, and measurements have a date. The tool it describes won’t hold still.
A stale A is worse than no grade, because someone is trusting it. Re-running the same test on a cadence is the only thing that keeps an index honest, and catches a rug pull when the tool surface changes.

Independent

Nobody can pay for a grade. No graded party gets review or approval rights.
A subscription changes what we watch and how often, never the letter we publish. The harness is open and deterministic, so any grade is something you can re-run and check, not take on faith.
03/How monitoring works

Re-grade, detect, alert.

01 · Re-grade

The whole index re-run on a schedule, the same behavioral test, repeated, not a one-time snapshot.

02 · Detect

Each run compared to the last. A drop, a new failing probe, or a fingerprint mismatch is flagged against what graded before.

03 · Alert

Your team hears about a regression before your users do, with the evidence bundle attached.

04/Funding

How polygraph gets funded.

Free to read, and not paid for by anyone we grade. Here’s where the money comes from instead.
$POLYGRAPH

The Bankr community launched $POLYGRAPH; we didn’t issue it. We claim the dev fees publicly and use them to fund the work: the harness, the grades, and the evidence stay free to read.

Nobody can pay for a grade. No graded party gets review or approval rights over their result.

View $POLYGRAPH on Bankr →

Not financial advice. The token funds the work; it doesn’t move a grade.

Put an independent trust layer under your ecosystem.

Tell us what your network ships. We'll stand up a continuous, independent trust index, graded, then re-graded on a cadence, and keep it current as you add to it.

Start monitoring See a live index

Prefer to write us directly? hello@polygraph.so