Skip to main content
Glama

Search the African Commission and Court (paragraph level)

banjul_search

Search African Commission and Court human rights decisions by paragraph, using full-text queries and filters to find cited violations, states, articles, judges, and cases with verbatim results.

Instructions

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-text search over paragraphs. Query syntax as on the website: words must all appear in one paragraph; "exact phrase"; OR between alternatives; NOT word or -word; prefix* (at least 3 letters); field operators violation:7, nonviolation:2, article:7, alleged:7 (Charter; other instruments as violation:ICCPR-14), state:Kenya, case:245/02 or case:005/2013, judge:"Ben Achour" (separate opinions by a judge), cites:155/96 or cites:"Alex Thomas" (decisions citing a decision), year:2015-2020, body:court. Accents and capitals are ignored. An empty query lists decisions (newest first) matching the filters. Returns, for each hit, a citation with the paragraph number, the VERBATIM paragraph, the decision metadata and findings, the doc_id for banjul_get_decision, and links to the PDF page and to the website.

Tool call template

This tool declares no required parameters. Add optional parameters from the input schema as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "banjul_search"
  }
}

Input Schema

NameRequiredDescriptionDefault
bodyNoOnly the African Commission or only the African Court.
sortNorel = relevance (default with words), new = newest first, old = oldest first, cited = most cited by other decisions first (leading cases).
typeNoKinds of decision (substring match), e.g. ["merits"], ["provisional measures"], ["admissibility"], ["separate opinion"].
groupNoparagraph = best paragraphs across decisions; decision = one entry per decision with its best paragraphs.paragraph
judgeNoSeparate opinions and declarations by a judge of the Court (surname), e.g. "Ben Achour", "Niyungeko", "Bensaoula".
limitNo
queryNoe.g. "free legal assistance" violation:7, indigen* "ancestral land", expulsion state:Angola.
stateNoRespondent State, e.g. Tanzania, Kenya, "Côte d'Ivoire".
offsetNo
articleNoAn article with a finding, e.g. "7" (African Charter), "ICCPR-14", "Maputo-4", "ACDEG-17".
findingNoWith article: violation found, no violation found, either, or alleged (claimed by the complainant or ruled on).any
year_toNo
year_fromNo
operative_onlyNoSearch only the operative parts.
include_opinionsNoInclude separate opinions of the Court.
include_translationsNoInclude the unofficial English machine translations of the 47 French-only Commission decisions.

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv1.0.0

TDQS

A4.5/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations, the description carries the behavioral burden and does it well: it discloses corpus boundaries, the OCR caveat with a verify-the-PDF instruction, what each hit contains (citation with paragraph number, verbatim paragraph, metadata/findings, doc_id, PDF and website links), and how an empty query behaves. It does not discuss result totals, relevance ranking, or performance limits, which keeps it below a 5.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Front-loaded with corpus then capability, and most sentences earn their place given a 16-parameter search tool with a custom query syntax. The query-syntax block is a dense run-on that would scan better as a list, and some of it (e.g. the prefix and field-operator examples) overlaps examples already in the query parameter schema.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

With no annotations and no output schema, the description still tells the agent what the corpus covers, how to phrase queries, what a hit contains, what an empty query does, and where to go next via doc_id. Pagination, sort defaults, and enum values are already documented in the schema, so nothing essential is missing for correct invocation.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters5/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

For 16 parameters at 75% schema coverage, the description adds real meaning beyond the schema: it spells out the query mini-language (all-words-in-one-paragraph, "exact phrase", OR, NOT/-word, prefix* with a 3-letter minimum) and maps field operators onto filters (violation:ICCPR-14, article:7, case:005/2013, judge:"Ben Achour", cites:155/96, year:2015-2020, body:court). It also notes accents and capitals are ignored, which the schema never says.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description names the exact corpus (African Commission library decisions plus African Court AJUDATA judgments/orders/separate opinions) and the unit of retrieval (numbered paragraphs), then states the operation: full-text search over paragraphs. That is a specific verb plus resource, and the closing note that hits carry a doc_id for banjul_get_decision implicitly separates this from the fetch sibling.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

It gives concrete operating context: an empty query lists decisions newest-first subject to filters, and hits link onward to banjul_get_decision. The 'check the PDF before quoting' warning is useful guidance for the OCR-corrupted portion. It stops short of explicitly routing the agent away from this tool to banjul_facets or banjul_check for counting/aggregation, so no true exclusions are stated.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.