immut for founders and AI agents

You build the company. Your agent builds the proof.

A pitch will only get you so far. When an investor or buyer considers your company, they ask whether your IP and compliance story holds up under diligence. Your agent decides which files matter, protects them as you create them, so the data room is harder to attack and easier to believe.

Use any agent: Claude · ChatGPT · Codex · Cursor
Install the agent skillBook a funding consultation
You agree the scope before anything is stored. Court-ready certificates on every proof.
terminal · your AI host
$ npx skills add enroh-ops/immut-agent
$ immut setup
short setup · scope agreed · you approve uploads
$ immut sweep
12 files proven · 3 skipped, not final · 0 silent actions
$

What investors and buyer see when diligence starts

Stand out by de-risking the investment decision.

Continuous proof of IP and compliance while you build. Not a last-week cleanup before the data room opens.

A data room that stands up to diligence.
Policies, security packs, and product artifacts dated when they were finalised, not when someone remembered to prepare.
immut · data-room
$ immut status data-room/series-a
security-policy-v3.pdfproven
access-review-q2.xlsxproven
architecture-overview.pdfproven
vendor-register.csvpending
IP that is harder to dispute.
Designs, code snapshots, research notes, and pitch materials proven as you ship, before every share with investors or partners.
immut · ip-trail
$ immut sweep --lane ip
model-v2-weights.hashproven
prototype-cad-rev12.stepproven
research-notes-2026-06.mdproven
3 dated proofs · court-ready certs written
Compliance that is not cleaned up later.
SOPs, risk assessments, and vendor docs proven at finalisation, so the diligence question "when did this exist?" has an answer.
immut · verify
$ immut verify risk-assessment-final.pdf
anchored 2026-03-14 09:21 UTC
match: file unchanged since proof
“when did this exist?” · answered
$
Protect before you disclose.
Prove the deck, model, and technical pack before the meeting. If ideas walk, you still hold dated proof of what you had.
immut · before the meeting
$ immut protect pitch-deck-v4.pdf --cert
pitch-deck-v4.pdf.immut.json written
certificate.pdf written
proven before disclose
$

A continuous proof portfolio from your agent, not a larger legal bill.

Skill · MCP · CLI · API

Four ways in. Build whatever the raise needs.

Same key, same proofs. Let your agent run the skill, connect over MCP, script the CLI, or wire the API into your own tools and workflows.

01
live
Agent skill
One install teaches Claude Code and compatible agents the whole job: setup, sweep, and store selected files on immut, organised for your objective.
$ npx skills add enroh-ops/immut-agent
$ immut setup
store file + proofPOST /api/v1/documents
organise foldersGET/POST /api/v1/folders
Install from GitHub
02
live
MCP server
Connect immut over MCP and the AI host you already use, Claude, ChatGPT, or Cursor, gets immut as native tools: protect, organise, and verify from inside the conversation.
$ claude mcp add immut
tools: protect · folders · certificates · verify
works withClaude · ChatGPT · Cursor
list workspacesGET /api/v1/workspaces
Set it up
03
live
CLI
Hash-only utility for engineers: fingerprint, prove, and verify without uploading bytes. Build it into CI, release scripts, or your own pipelines.
$ immut proof create --file deck.pdf
$ immut verify txHash --file deck.pdf
hash-only proofPOST /api/v1/proofs
public verify, no keyGET /api/public/verify
CLI on GitHub
04
live
REST API
A deliberately small surface for custom integrations: store documents, organise folders, pull certificates. 60 req/min, 10,000/day per key.
POST /api/v1/documents
multipart: file + workspace + folder
certificate PDFGET /api/v1/certificates
machine-readable docsGET /api/v1/docs
Read the quickstart
Full docs: quickstart · agent playbook · OpenAPI

Use cases

Just ask the AI you already use.

These are not commands to learn. They are the questions founders type into Claude, ChatGPT, or Cursor once immut is connected.

YOU ASK
“Protect this file.”
→ proven, certificate attached, filed in the data room
YOU ASK
“How many files has the team protected this month?”
→ 38 this month · 4 pending · by folder and by person
YOU ASK
“Show me every IP file we've saved in the last month.”
→ 11 files, each with its dated proof and folder
YOU ASK
“Share the blockchain link for every board meeting in the last 12 months.”
→ 12 board packs, one verification link each, ready to forward
YOU ASK
“Which emails are we relying on as evidence that we should be documenting properly?”
→ 6 threads flagged, each with a suggested file to prove instead
YOU ASK
“Is the data room ready for diligence? What's missing?”
→ coverage by folder, plus the gaps a diligence analyst would find

Two ways to run it

Run it yourself, or have us run it.

The hard part is not the certificate. It is deciding which files need one, every week. That is the agent's job. How it gets set up is up to you.

Free and open source
The agent skill
For teams already running AI agents. Install the skill, answer a few questions once, and it decides what to protect from then on. It runs on your own AI host, on a daily recurring check. The skill is free. You pay for immut storage and proofs.
Install the agent skill
Managed
We build and maintain it for you
For teams who will not wire up agents and connectors themselves. We implement it: objective, folder structure, connectors on your host, a first pass you approve, then go live. Optional monthly management keeps it deciding for you as connectors break and scope changes.

One-time implementation, optional monthly management, plus your immut subscription.

Your boundaries

Low-trust integration. High-trust diligence.

Giving an agent an immut key is a controlled decision. The proof stays independently verifiable for parties that do not trust your internal systems.

Never expands scope.
It sees the folders you configured and nothing else.
Never touches your billing.
Keys are scoped to documents, folders, and certificates.
Never follows instructions found inside your documents.
Document content is data, not commands.
Never acts silently.
Every proof and every skipped candidate is reported back to you.
api key · what it can do
imut_live_••••••••
documents:writegranted
folders:writegranted
certificates:readgranted
billingno access
file contentsyou approve each one
Classification happens on your machine first. You approve what the agent may store.

Show up to diligence with proof, not promises.

Install the skill, point it at the folders that matter, and build a fundraise-ready proof trail while you ship.

Install the agent skillBook a funding consultation
$ npx skills add enroh-ops/immut-agent