Intelligence MCP
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
Ratiqora Intelligence MCP — Alpha 4
Ratiqora helps MCP developers trace a conclusion back to the observations that support it. The local server checks evidence integrity and graph references, and exposes redacted evidence metadata over stdio.
Download the one-minute demo and open the HTML file in your browser. No installation or Python is needed. The English walkthrough asks whether a product team should lower its photo upload limit. Compare three reports, choose a next step, and optionally explore the evidence trail.
A two-minute example: photo uploads
Three customers reported failed uploads for photos above 10 MB. Another uploaded a 2 MB photo successfully. These observations make file size worth testing, but they do not establish a 10 MB limit.
Trace step | Synthetic example |
Insight | Compare upload outcomes before changing the form. |
Finding | The failed uploads were larger than the successful one; format and other causes remain untested. |
Item | The reports associate larger files with failures in this small sample. |
Evidence | The reports are linked by ID and checked against the SHA-256 of their exact bytes. The raw payload is not returned. |
Source | A synthetic batch of customer feedback. |
This table is an illustrative scenario. The packaged trial builds its own synthetic graph. A valid graph has intact references; it does not prove the conclusion or authenticate a source.
The walkthrough’s three-report scenario can be turned into a one-page decision brief with supporting and conflicting reports, open questions, and a next step.
Related MCP server: io.github.BrienBounded/evr
Build your own decision brief
Run the local Decision Workbench to link your observations to public or fictional source references. Ratiqora checks the internal links before you can download a Markdown brief. You provide the finding and evidence roles.
Try it
Download the two wheels, constraints and checksum file from Alpha 4.
Follow the quickstart for Windows or Linux.
Run the synthetic trial. To explore the graph visually, generate the local trace viewer and open the resulting HTML in a browser. Add
--conflicting-uploadsto explore two equally large photos with opposite outcomes. The generated page needs only a browser. No model API key is needed.
Alpha 4 is an experimental single-project, in-memory runtime. Use synthetic data for these examples. It is not a hosted service, a general file connector or a production system.
Contribute
Demo feedback: Tell us what was clear and where you got stuck. This opens a public GitHub issue draft.
Docs and packaging: Test the quickstart on a fresh Windows or Linux environment and report the first step that could be clearer.
MCP examples: Propose a checkout scenario with conflicting reports and show what conclusion the evidence supports.
Write UI text, documentation, examples, and test fixtures in English. Use synthetic data and include reproduction steps with a focused PR. Never include real customer data or raw evidence.
This repository is a curated snapshot of the MCP source. The matching Core wheel includes its Python implementation. See LICENSE, NOTICE and SECURITY.md.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Maintenance
Related MCP Connectors
Physical-world evidence and operability checks with provenance and explicit data gaps.
Read-only verifier for 25 ProofRelay MCP tools and non-confidential evidence bundles.
Evidence-readiness MCP server: validate, audit, and score briefs, memos, and evidence packs.
Evidence-bound second-opinion audit of an agent conclusion against caller-supplied evidence.
Related MCP Servers
- AlicenseAqualityBmaintenanceA local-first MCP server for retrieving a small evidence set and recording reviewed conclusions, policy-gated and redacted without giving an agent general filesystem access.5MIT
- AlicenseAqualityDmaintenanceEnables agents to check whether an execution trace contains the evidence needed for post-hoc causal reconstruction, retrieve individual requirement details, and list evidence requirements per failure class over MCP stdio. It works on files, directories, or inline JSON locally and never transmits or persists data.3Apache 2.0
- FlicenseNot gradedqualityAmaintenanceEnables agents to gather bounded, reproducible runtime evidence by coordinating local profilers, benchmark tools, and trace processors against explicit artifacts or captured commands. It exposes typed analysis and capture operations over MCP that return provenance-tracked results inline, with optional durable preservation and replay of prior evidence.-
- AlicenseNot gradedqualityCmaintenanceEnables agents to validate and deterministically compare backtest, replay, and live pipeline traces, inspect the first observable divergence, and investigate the supplied evidence through a local read-only interface.MIT