Skip to main content
Glama

mcp-banjul

Model Context Protocol server for Banjul Researcher (https://lszoszk.github.io/banjul-researcher/): paragraph-level search over the decisions of the African Commission on Human and Peoples' Rights and the judgments, rulings, orders and separate opinions of the African Court on Human and Peoples' Rights, with each paragraph's case name, communication or application number, date and paragraph number, plus a citation check.

The server loads the same data as the website (docs.json and the paragraph files published on GitHub Pages) and runs the same search, so an assistant and the website return the same results. Every result is the verbatim text of the decision, so answers stay citable to the paragraph.

Tools

Tool

Purpose

banjul_search

Paragraph-level full-text search: phrases, OR, NOT, prefix*, violation:7, article:ICCPR-14, state:, case:, judge:, cites:, year:, body:; sort by relevance, date or most cited; filters for body, State, judge (separate opinions), years, kind of decision, article and finding, operative part only, separate opinions, machine translations

banjul_get_decision

Metadata, extracted findings, separate opinions, the decisions it cites and that cite it, and the verbatim paragraphs of one decision (by paragraph range or the operative part)

banjul_check

Finds the citations of the Commission and the Court in a text and checks the number, the name, the pinpointed paragraph and the quotation

banjul_facets

Corpus size and the values the filters accept, including the most cited decisions and the judges who wrote separate opinions

Related MCP server: HUDOC MCP

Hosted connector (nothing to install)

https://150.254.115.204/banjul-mcp/mcp – Streamable HTTP, no token.

  • Claude (claude.ai / Desktop): Settings → Connectors → Add custom connector → paste the address.

  • Claude Code: claude mcp add --transport http banjul https://150.254.115.204/banjul-mcp/mcp -s user

Run it yourself

npm install
node index.js          # stdio, for Claude Desktop or Claude Code
node http.js           # Streamable HTTP on 127.0.0.1:8015/mcp
npm test               # smoke test against the published data

DATA_BASE points to another copy of the published data; REFRESH_HOURS (default 6) sets how often it is reloaded.

Notes

Scanned decisions were read by OCR, and the findings of violation are extracted automatically from the operative parts: read the paragraph and the PDF before relying on either. The machine translations of French-only Commission decisions are unofficial and are left out unless include_translations is set.

Licence: PolyForm Noncommercial 1.0.0. The decisions are the work of the African Commission and the African Court; this project is not affiliated with or endorsed by either body.

Available Tools

4 tools
banjul_checkCheck citations of the African Commission and Court in a textA

Corpus: decisions of the African Commission on Human and Peoples' Rights (from its library) and English judgments, rulings, orders and separate opinions of the African Court on Human and Peoples' Rights (from AJUDATA), split into their numbered paragraphs. Scanned decisions were read by OCR: check the PDF before quoting. Finds the citations in a text (communication numbers "Communication 245/02", application numbers "Application No. 005/2013", names "X v State", "X, cited above") and checks each: does the number exist and match the name, which decision of the case is meant (by date, year or kind), does the pinpointed paragraph exist, and does a quotation next to the citation match the decision word for word. Status per citation: ok, warn (check it), bad (not found), amb (ambiguous), out (outside this database, e.g. ECOWAS Court). Use it on your own draft answer before giving it.

Inputs: text (required), compare_quotes.

banjul_facetsCorpus figures and filter valuesA

Corpus: decisions of the African Commission on Human and Peoples' Rights (from its library) and English judgments, rulings, orders and separate opinions of the African Court on Human and Peoples' Rights (from AJUDATA), split into their numbered paragraphs. Scanned decisions were read by OCR: check the PDF before quoting. Corpus size and the values the filters of banjul_search accept: bodies, kinds of decision, respondent States, articles with findings (with counts), the most cited decisions, judges who wrote separate opinions, years.

Inputs: no named parameters.

banjul_get_decisionGet a decision: metadata, findings and verbatim paragraphsA

Corpus: decisions of the African Commission on Human and Peoples' Rights (from its library) and English judgments, rulings, orders and separate opinions of the African Court on Human and Peoples' Rights (from AJUDATA), split into their numbered paragraphs. Scanned decisions were read by OCR: check the PDF before quoting. Full record of one decision by doc_id (from banjul_search): metadata, findings, separate opinions or the decision an opinion belongs to, the decisions it cites and that cite it (with paragraphs), and the VERBATIM paragraphs, optionally a range of paragraph numbers or the operative part only. Long decisions are returned in pages of at most 80 paragraphs: use from_para.

Inputs: doc_id (required), to_para, from_para, operative_only.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A4.2/5.0

    Scored across 4 tools

    Disambiguation5/5

    Each tool occupies a distinct role: banjul_search retrieves paragraphs, banjul_facets exposes corpus dimensions and accepted filter values, banjul_get_decision fetches a full record by doc_id, and banjul_check validates citations in a draft. There is no plausible way for an agent to confuse these operations, and the descriptions reinforce the boundaries.

    Naming Consistency4/5

    All names use a single banjul_ prefix with snake_case (search, facets, get_decision, check), which is predictable and readable. The only minor deviation is that 'facets' is a bare noun rather than a verb_noun construction, though it still clearly denotes corpus dimension discovery.

    Tool Count5/5

    Four tools is a tight, well-scoped surface for a legal-corpus search server: discover filters, search, retrieve a record, and verify citations. Each tool earns its place, and there is no redundant or filler tool.

    Completeness5/5

    The lifecycle is fully covered: banjul_facets reveals filter values, banjul_search finds paragraphs with rich filters, banjul_get_decision provides metadata, findings, citations, and verbatim paragraphs with pagination, and banjul_check closes the loop by verifying citations and quotations in draft answers. Empty-query listing and citation-graph data in the record retrieval leave no obvious dead ends.

    Maintenance

    ActivityMaintained
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • F
      license
      Not graded
      quality
      C
      maintenance
      Enables semantic search, outcome prediction, and legal document drafting across 67M+ Brazilian court decisions from 55 tribunals, with tools for jurimetric analysis and citation verification.
      -
    • A
      license
      Not graded
      quality
      B
      maintenance
      Enables AI agents to search and retrieve European Court of Human Rights case law by free text, Convention Article, respondent state, collection, importance, and date, then read full judgment texts by item ID.
      107 npm
      MIT
    • A
      license
      B
      quality
      B
      maintenance
      Enables citation-safe legal research and document drafting over official Turkish case law and legislation, searching a local 11-million-decision corpus with full-text and semantic retrieval or querying live sources like Bedesten, AYM, Danıştay and mevzuat.gov.tr. It also supports citation verification, petition and bundle preparation, change reports, privacy redaction, and source health checks through core and loadable extended tool sets.
      11
      4
      AGPL 3.0
    • A
      license
      A
      quality
      B
      maintenance
      Enables searching and reading Peruvian Constitutional Court jurisprudence by querying both the current thematic systematization and the 2001-2018 historical resolutions inventory at once, returning them fused and deduplicated as a single source. Supports full-text search, exact filters (case number, parties, dates, chamber, district, type, keywords), thesaurus and thematic browsing, and full metadata plus complete reasoning text with public PDF links, without credentials or browser automation.
      9
      2
      Apache 2.0