Deterministic research state, evidence validation and statistical recomputation.
Quick Install
Automated & IDE SetupCopy the AI prompt to install this server into Claude Code, Cursor, or another agent — or use 1-click editor setup below.
One-click editor setup isn’t available for this listing yet — we don’t have a confirmed install command, and we’d rather show nothing than point your editor at the wrong package or host. Follow the project’s own setup instructions, linked above.
Manual Client & Custom JSON ConfigExpand JSON ▾
No confirmed setup config for this listing yet. We only publish a config block when the install details come from the project itself — its README, its docs, or a verified owner. We haven’t found those for Academic Research Kernel, and we’d rather show nothing than a guess you’d paste into your client. Follow the project’s own setup instructions for the current steps.
Documentation Overview
academic-research-kernel
English · 简体中文 · 繁體中文 · 日本語 · 한국어 · Deutsch · Français · Español · Português · Русский · Bahasa Indonesia · Italiano · हिन्दी · العربية · বাংলা · اردو · Tiếng Việt · Türkçe · فارسی · Kiswahili · Polski
Harness-neutral deterministic research-state kernel for agentic research workflows. It unifies research object identity, empirical evidence receipts, tripartite claim-evidence graphs (CEG), and append-only research decision logs through a deterministic ingestion bridge and universal Model Context Protocol (MCP) server. 13 specialized scholarly skills act as producers and consumers of verified research state, with native support for Claude Code, Cursor, Codex, Gemini CLI, and Hermes Agent. Validation scope and external-service limitations are recorded in the audit.
Author: Junfu Shi (SJF, xngg1021), Hermes Agent. Current scoped offer: Source Lineage License 1.0.
Quick start
After 2.0.0 is published to PyPI, run with uv:
Publication is a separate release step: check the 2.0.0 release
manifest for the actual PyPI status. Before publication, download the release wheel
and use uvx --from ./academic_research_kernel-2.0.0-py3-none-any.whl academic-research-kernel mcp.
For a fixed product identity, use uvx academic-research-kernel@2.0.0 mcp.
Python 3.10–3.14 is supported; uvx --python 3.12 explicitly selects a supported interpreter.
All 12 deterministic tools, including statistics, are included in the base install.
Verify
The offline diagnostic checks dependencies, schemas and all 12 tool definitions.
Optional services can remain unconfigured. Use doctor --json for machine-readable output.
Package install
Alternatively, install with pipx install academic-research-kernel==2.0.0 or
python -m pip install academic-research-kernel==2.0.0 inside a supported virtual environment.
[full] adds libraries for extended scientific skill examples; [qa] adds repository
validation tools. GPU backends and paid services remain optional.
Hermes
Install uv so it is on Hermes' process PATH, then install and explicitly enable the portable plugin:
The existing academic-skills identity and all 13 skills are preserved. The plugin's
mcp.json invokes uv run --frozen --no-dev --no-editable with Python 3.12 and the
checkout's uv.lock; its environment lives under ${PLUGIN_DATA}/runtime.
It bootstraps dependencies on first use, without borrowing Hermes' or the system's Python
packages. First use needs network access to obtain missing packages/interpreter.
The older skill-only tap remains supported under Integration & Portable Usage.
From source
See distribution and release for dependency classification, reproduction, release identity, publication checkpoints and validation scope.
License
The snapshot containing this notice adopts Source Lineage License 1.0 for the Covered Material and rights identified in LICENSE-APPLICATION.md. The first SLL commit and tree, and the later boundary-recording commit, are distinguished in LICENSE-HISTORY.md and SOURCE-LINEAGE.md. Snapshots from that recorded transition forward retaining this notice carry the same scoped offer.
Historical snapshots through 439ce4a29d6d2d9137308d7f9f045863e9f0c5b3 were MIT-licensed, subject to the terms applicable to those copies. Recipients retain valid MIT permissions and need not migrate to SLL. The former project MIT text is preserved; tests/upstream/LICENSE and its third-party provenance remain unchanged. The new root offer does not erase those rights.
SLL broadly permits use, study, modification, commercial use, distribution and proprietary additions, subject to applicable license, notice and source-lineage conditions. It is not copyleft and requires no source disclosure. Pure network service without supplying copies does not by itself trigger the Core service-lineage notice condition. There is no express patent grant. Third-party material remains under its own terms. The exact English LICENSE controls this informational summary; LicenseRef-Source-Lineage-1.0 is a local reference, not SPDX assignment, and no OSI approval is claimed. Contribution intake is separate from downstream license permissions.
Skills
| Skill | Version | What it does |
|---|---|---|
skills/academic-source-verification | 1.2.0 | Cross-check identity and source-specific citation counts; inspect update/retraction signals; locate OA text and verify PDF identity |
skills/literature-analysis | 1.3.0 | Twelve workflows: topic similarity, local text overlap, counter-evidence, author profiles, mock review, fallacy checks, review matrix, journal candidates, BibTeX, bilingual reading, research-gap screening, reproduction |
skills/academic-writing | 1.1.1 | Editing, citation guidance (ISO 690, APA, MLA, Chicago, IEEE, AMA, and regional profiles), journal instructions, optional detection services, submission materials, and institutional requirements |
skills/math-computation | 1.2.1 | Existing domain/task routing with corrected numerical/statistical examples; four domain/advanced reference files |
skills/quantitative-paper-audit | 1.1.0 | Recompute reported statistics (effect size, p values, CIs, OR/RR, achieved power) and detect numerical mismatches |
skills/research-reproducibility | 1.0.1 | Fourteen-stage reproduction audit pipeline with structured verification checklists, five fact tiers, and reproducible audit records |
skills/systematic-review-meta-analysis | 1.0.1 | PRISMA search logs, literature screening logs, effect-size conversion, heterogeneity, fixed/random pooling, sensitivity and publication-bias diagnostics |
skills/literature-watch | 1.1.0 | Weekly blueprint: watch topics, authors and DOI citing works on OpenAlex and Crossref; deduplicate and report only new items |
skills/retraction-watch | 1.1.0 | Weekly blueprint: recheck a DOI watchlist against OpenAlex is_retracted and Crossref update records (update-to signals); report only status changes |
skills/research-object-identity | 1.1.0 | Deterministic research-resource identification and provenance tracking: identifier normalization, 5-state verification, content-addressed derivation graphs, causal DAG validation, and sub-100ms lineage tracing |
skills/claim-evidence-graph | 1.0.0 | Deterministic scientific claim–evidence linking connecting assertions, evidence records, and computational provenance |
skills/decision-ledger | 1.0.0 | Research Decision Log: deterministic, append-only log of research decisions, failed attempts (negative results), reasons for stopping routes, and outcome revisions |
skills/cross-review-five | 2.0.0 | Dynamic multi-reviewer panel orchestration supporting arbitrary models/subagents (Kimi K3, DeepSeek V4 Pro, GLM 5.3, Claude, Gemini, etc.): v2 Sparse Deliberation pipeline with Kuhn-Munkres Hungarian assignment, assertion-level clustering, targeted anonymous challenge, and P0-P3 severity grading |
There are 21 Markdown reference files across the thirteen skills. References load only when needed.
Scholarly Standards & Multi-Profile Baseline
Citation styles, reporting criteria, and metadata contracts depend on the target journal, institution, funder, discipline, and jurisdiction. The repository establishes ISO 690:2021 (Bibliographic references), ISO 5127:2017 (Information and documentation vocabulary), and W3C PROV (Provenance data model) as international baselines, alongside regional profiles (e.g., GB/T 7714-2025 in Mainland China, UNE-ISO 690:2024 in Spain, DIN ISO 690:2021 in Germany) and disciplinary standards (APA 7th, IEEE, ACM, Vancouver, Chicago, PRISMA 2020, ICMJE). Target venue requirements take precedence over default profiles. See the Scholarly Standards Architecture and Natural Terminology Guide.
The final PR #12 repair validates DOI/identifier coherence, inverse quantitative verdict flags, and the shared kernel/receipt uncertainty contract. Cache completeness is reconstructed from retained source envelopes and normalized bindings, independently of the receipt's mutation lists; removing both a record and its cache declaration cannot produce an accepted replay. See the four-blocker closeout.
Integration & Portable Usage
Related MCP Servers
Reviews
No reviews yet — be the first to share how this listing worked for you.
Frequently Asked Questions about Academic Research Kernel
AllMCPs Directory Badge
Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.
Badge Style:
Live Dynamic SVG Preview
Markdown (GitHub README)
[](https://allmcps.com/mcp/academic-research-kernel)HTML Embed
<a href="https://allmcps.com/mcp/academic-research-kernel"><img src="https://allmcps.com/api/badge/academic-research-kernel?style=directory" alt="Academic Research Kernel on AllMCPs" /></a>