formal-proof-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| lean_checkA | Compile a Lean 4 snippet and return the compiler's verdict. On failure the real error is returned so the agent can repair its own proof. If no Lean toolchain is installed the result is |
| axiom_auditA | Audit |
| boundB | What a k-of-n record actually supports, as an exact one-sided Clopper-Pearson bound. 'It passed every time' is not a bound. |
| gridlock_checkA | Certify that a wait-for relation cannot wedge. Returns the actual cycle when one exists. Supply ranks to also check strict decrease, which is what bounds progress — acyclicity alone only forbids a cycle. |
| cert_verifyC | Verify a signoff-cert/v1 certificate: digests, gate consistency, and the false-pass bound RECOMPUTED from the evidence. Requires |
| residency_checkB | Can a cross-tenant cache probe on this model even be interpreted? Returns |
| prereg_checkA | Before running an experiment, check that its decision rule CAN come out both ways. Returns UNFALSIFIABLE when the finding -- or the null -- can never fire over the declared metric supports. Requires |
| state_floorC | How many states must a system distinguish to answer a question about its past? An exact count over an enumerated situation space. Optionally proves a state budget cannot meet it. Requires |
| gate_countA | Exactly how many states does removing a check admit? Replaces 'we found no escapes' with a count. Zero means the check is REDUNDANT, which is a finding. Requires |
| evidence_auditA | Run every applicable verifier over a directory and aggregate to ONE verdict -- the weakest leg, never the mean. Requires |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |