zeroH MCP
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| ZEROH_API_KEY | No | API key for the zeroH Enterprise Sovereign Enclave. Optional; when configured with ZEROH_API_URL, enables Tier 2 remote enclave verification. Empty value runs in zero-config local mode. | |
| ZEROH_API_URL | No | URL of the remote zeroH Sovereign Enclave. Optional; when configured with ZEROH_API_KEY, enables Tier 2 remote enclave verification. Empty value runs in zero-config local mode. |
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 |
|---|---|
| validate_claimA | Intercept and validate a factual, mathematical, or scientific claim before presenting it. Returns Coherence Index (0.0-1.0), Confidence Score (0%-100%), and Deterministic Status (VERIFIED_COMPUTED_PASS, REJECTED_HALLUCINATION, BOUNDARY_BREACH, or NEEDS_REVIEW). |
| verify_stepA | Submit a single reasoning or mathematical step to the tamper-evident SHA-256 session ledger before stating it. Supported step kinds: 'compute' {expr, value}, 'solve' {equation, claimed: {x: val}}, 'reasoning' {text}, 'interpretation' {text}. |
| session_reportA | Seal and audit all steps submitted in a reasoning session. Returns FULLY_PROVEN only if every active step was deterministically proven, along with the SHA-256 Merkle head hash. |
| computeB | Compute the exact rational fraction and decimal value of a numeric expression using zero-drift AST rational arithmetic. |
| solve_equationB | Solve an algebraic equation and verify every solution by exact substitution. |
| intercept_tokenC | Query the Sovereign Enclave canonical ledger for a token or root reference. Requires an authenticated Sovereign Enclave connection (ZEROH_API_URL). |
| zk_verify_receiptC | Generate a cryptographic commitment receipt over a verified claim (or request an R1CS proof from a connected Sovereign Enclave). |
| epistemic_consensusC | Byzantine Fault Tolerant (BFT) Multi-Agent Epistemic Consensus Gate. Evaluates agent proposals across the 4-tier epistemic lattice (QATI, ZANNI_GHALIB, SHAKK, WAHM) with automatic hallucination veto. |
| enclave_statusA | Check the health, execution tier, and IP protection status of the zeroH MCP server and Sovereign Enclave bridge. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 9 tools
The tools split into recognizable clusters (validate_claim / verify_step / session_report form a pipeline; compute and solve_equation are clearly distinct math operations), but several tools sit in an overlapping 'verify things' space. validate_claim, epistemic_consensus, and zk_verify_receipt all evaluate or attest a claim, and an agent could plausibly pick the wrong one without careful reading. intercept_token and enclave_status are also both enclave-facing lookups that could be confused.
All names use snake_case, which is good, but the verb conventions are mixed: verb_noun (validate_claim, verify_step, solve_equation, intercept_token), bare verb (compute), noun_noun (session_report, enclave_status, epistemic_consensus), and a prefixed form (zk_verify_receipt). It stays readable, but there is no single predictable pattern to infer a new tool's name from.
Nine tools is a well-scoped set for an epistemic-verification server, with each tool covering a discrete capability (arithmetic, solving, step logging, session sealing, receipt, consensus, health). Two tools (intercept_token, zk_verify_receipt) depend on an optional Sovereign Enclave connection, which makes them conditional rather than redundant.
The core lifecycle is covered end-to-end: validate a claim, log individual steps, seal and audit the session, then emit a cryptographic receipt, plus math primitives and a health check. Minor gaps exist — no way to list/clear session steps or start/reset a session, and no batch validation — but agents can work around these.