Cognitive Fab
Thirteen tools · one gate
Tools Services Blog Get started GitHub ↗

Tools, grouped by
what you came to do

Each one is a plugin to a coding agent, and each runs a deterministic check — no model call at check time. Start where your problem is: the code an agent writes, what an agent is allowed to do, or the outcome you are designing toward.

Verify the code

The SAM v2 gate

Six tools in one lifecycle around one deterministic gate — they run in order, and each hands the next an artifact rather than a claim — and a seventh that draws what any of them produced.

polynv→polygen→polygraph→polygate→polyrun→polyvers
polynvelicitDesign The invariants a system must never break — harvested from the machine's own vocabulary, graded by mutation, confirmed by a human rather than invented by a model. open source polygenauthorDesign Authors new state machines that are checkable from the first line — contract and implementation drafted together, self-repaired, shipped with a regression corpus. open source polygraphauditVerify The verification engine. Explores every reachable state over declared domains and returns the shortest action sequence that breaks your rules. open source polygategateVerify The CI/CD merge gate. Keyless replay, model checking and compatibility gates on every merge request, so unverified changes never reach a protected branch. open source polyrunexecuteOperate Durable execution for verified machines — one commit per step, effects emitted exactly once, and a gate on the check-then-effect sequence itself. open source polyversevolveEvolve Checks whether a version change is safe against the live fleet before it deploys — classified, seeded against real in-flight state, gated on the verdict. open source polyvizdrawevery stage Renders verification artifacts as diagrams. Same inputs, byte-identical SVG, no model call at render time. open source
Govern the agent

What an agent is allowed to do

The same substrate pointed at a different question: not “is this call acceptable?” but “can this session ever reach a situation it must never be in?” The argument, the tools and the reading in one place →

polycheckcheck agent policyOperate Reads your .claude settings and MCP config and determines whether your policy allows a path from untrusted input to credential or source egress. open source polyflowrun agent workflowsOperate A workflow engine for AI agents. Admits a workflow only if it model-checks, then runs it durably and hands the agent one work order at a time. open source polycrewcrew a runOperate Several agents — and the people among them — on one polyflow run at the same time: a shared broker, orders one of them can claim, no order done twice, and a journal that says who did what. open source polysecmandate + red linesVerify Treats a leak as a reachable state rather than a clever prompt, and checks the guardrail on every path before it ships. engine · not distributed polymanenforce policyOperate Sits between an agent and its tools, tracks what the session has touched, and refuses the call that would complete the leak — before it runs. engine · not distributed
Design the outcome

Before anyone writes code

Cartographdesign the outcomeDesign A browser-local designer for goal-driven systems — four diagrams over one shared model, built on the BOLT strategy notation. free · in-browser

Not a tool — Polytrion

Our digital-twin design platform is a separate line of work, and it is not something you install. It runs as a hosted platform in soft launch, by invitation. See what it does →