The Mycelitree verification protocol.

Mycelitree measures the distance between what an AI output claims and what a grounded reading supports. It's not a guardrail. It's a measurement layer — the thing you consult before you treat a model's output as a decision.

What this is

Mycelitree sorts each claim by how grounded it is in verifiable source material, then flags the patterns that make ungrounded output read as confident. Full protocol documentation is coming soon.

Honest scope

This is a heuristic protocol. It does not detect hallucinations with ML-level sophistication. It detects patterns associated with unsourced, inflated, or performed output. False positives occur on rigorous prose that happens to be dense. False negatives occur on confident fabrication that imitates grounded style. Use it as one input to review, not as a verdict.

Who it's for

Engineers building LLM applications who need auditable output checks
Enterprise teams deploying AI in regulated workflows
Researchers measuring model behavior across sessions
Anyone who wants their AI outputs to come with receipts

Quickstart

Quickstart.

Run an audit on a piece of text in under a minute.

1. Install the CLI

Terminal

lorem install --placeholder ipsum-dolor-sit-amet

2. Run your first audit

Terminal

placeholder lorem "consectetur adipiscing elit sed do eiusmod,
tempor incididunt ut labore et dolore magna aliqua..."

Output

{
  "lorem": 0.00,
  "ipsum": "placeholder",
  "dolor": {
    "sit": 0.00,
    "amet": 0.00,
    "consectetur": 0.00
  },
  "adipiscing": "placeholder",
  "elit": "placeholder eiusmod tempor incididunt"
}

3. Or pipe from a file

lorem ipsum.txt | placeholder dolor --sit-amet

Introduction MCP Install

Install the MCP server.

Model Context Protocol server that exposes Mycelitree's audit tools to any MCP-compatible client.

Install

lorem install --placeholder consectetur-adipiscing-elit

Verify

lorem-ipsum --placeholder
# lorem-ipsum 0.0.0

Quickstart