Other formats agents might prefer:
markdownjsonllms.txt

Agent? You probably want markdown or json, or Pod over MCP.

Reported issues for mcp-reunion

Pod holds 16 of 25 problems reported by people outside the maintainer team. Issues filed by the project's own owners, members and collaborators are excluded entirely — a maintainer's release checklist is not a warning to a prospective user.

Back to mcp-reunion.

Most discussed

Add examples directory with prompt recipes

Context

Users need concrete prompts to understand what the MCP can do. Examples also help directories like Glama and GitHub visitors evaluate the project quickly.

Proposed files

  • examples/claude-desktop.md
  • examples/civic-analysis.md
  • examples/tourism.md
  • examples/elections.md
  • examples/transport.md

Acceptance criteria

  • Each example includes a user prompt and expected tool flow.
  • Examples avoid hardcoded fragile numbers unless sourced from a smoke-tested query.
  • READ

Read the thread · 2026-08-19 · closed · 1 comment

Add optional local-only tool usage analytics

Context

It would be useful to know which tools are actually used during local development and demos, without sending telemetry to any server.

Proposal

Add opt-in local logging controlled by an environment variable such as MCP_REUNION_TELEMETRY=local.

Acceptance criteria

  • Disabled by default.
  • Logs only local file data, never remote telemetry.
  • Captures tool name, timestamp, duration, success/failure, and upstream dataset ID if applicable.
  • README documents privacy behavior clear

Read the thread · 2026-08-19 · closed · 1 comment

Add release automation for npm, GitHub Release, and MCP Registry

Context

Releasing currently involves several manual steps: version bump, npm publish, GitHub Release, MCP Registry, and Glama sync/release.

Proposal

Add a release workflow that standardizes the process and reduces mistakes.

Acceptance criteria

  • Workflow can be manually dispatched with version input.
  • Runs build, tests, smoke, and audit before publishing.
  • Publishes npm only from a clean tag.
  • Creates GitHub Release notes.
  • Publishes/updates MCP Registry metadata.
  • Documents requ

Read the thread · 2026-08-19 · closed · 1 comment

Improve transport module with GTFS trip-planning helpers

Context

The transport module exposes routes/stops, but users naturally ask route-planning questions: how do I get from A to B, which lines serve a stop, and what alternatives exist?

Proposal

Add lightweight GTFS helpers around Car Jaune/local datasets.

Candidate tools

  • reunion_find_nearest_transit_stops
  • reunion_list_routes_serving_stop
  • reunion_estimate_transit_access_between_communes

Acceptance criteria

  • Do not overpromise realtime planning unless realtime data exists.

Read the thread · 2026-08-19 · closed · 1 comment

Add location-radius search helpers

Context

Many user questions are spatial: nearby schools, health professionals, sport facilities, bus stops, museums, etc. Current tools mostly filter by commune or text.

Proposal

Add helpers that accept coordinates and radius in meters, then return nearby records from selected datasets.

Candidate tools

  • reunion_nearby_schools
  • reunion_nearby_health_professionals
  • reunion_nearby_sport_facilities
  • reunion_nearby_car_jaune_stops
  • reunion_nearby_museums

Acceptance crite

Read the thread · 2026-08-19 · closed · 1 comment

Promote popular catalog datasets into dedicated modules

Context

The catalog tools can query ~270 datasets, but popular datasets deserve dedicated tools with curated schemas and safer parameters.

Candidate areas

  • volcanology / Piton de la Fournaise
  • natural risks
  • water and sanitation
  • agriculture
  • energy
  • local finance / fiscality if available

Acceptance criteria

  • Identify 5-10 high-value catalog datasets.
  • Add at least one new dedicated module or extend existing modules.
  • Add smoke tests for new datasets.
  • Update README module c

Read the thread · 2026-08-19 · closed · 1 comment

Generate a dataset coverage report

Context

The README says the MCP covers 99 tools across 22 modules, but users cannot easily see which datasets are covered and which remain catalog-only.

Proposal

Generate COVERAGE.md with:

  • module name
  • dataset ID
  • tool names
  • source portal
  • smoke status
  • whether coverage is dedicated or generic catalog-only

Acceptance criteria

  • Add script npm run docs:coverage.
  • Generated COVERAGE.md is deterministic.
  • README links to COVERAGE.md.
  • CI verifies the report is up to da

Read the thread · 2026-08-19 · closed · 1 comment

Add live schema drift detection for upstream datasets

Context

The BAN dataset recently changed field names and broke the smoke test. A schema drift detector would catch this kind of upstream change earlier and make fixes faster.

Proposal

Create a script that checks expected fields for every dedicated dataset and reports missing/renamed fields.

Acceptance criteria

  • Add a manifest of expected dataset fields used by tools.
  • Add npm run check:schema.
  • CI scheduled job runs weekly.
  • Failure output clearly lists dataset ID, missing field

Read the thread · 2026-08-19 · closed · 1 comment

Most recent

Add a natural-language router for relevant tools and datasets

Context

With 99 tools, agents may not always pick the best tool. A router tool can help map a user question to relevant tools and datasets.

Proposal

Add reunion_find_relevant_tools that accepts a user question and returns:

  • suggested dedicated tools
  • candidate catalog datasets
  • recommended query flow
  • caveats / missing coverage

Acceptance criteria

  • Uses deterministic matching first: tool names, descriptions, module keywords, dataset metadata.
  • Does not call an LLM or require A

Read the thread · 2026-08-19 · closed · 1 comment

Add high-level insight tools for common civic questions

Context

The MCP already exposes many dataset-level tools. The next usability jump is higher-level tools that combine datasets around real user questions.

Candidate tools

  • reunion_commune_risk_profile
  • reunion_tourism_dashboard
  • reunion_transport_accessibility_profile
  • reunion_public_services_profile
  • reunion_housing_pressure_profile

Acceptance criteria

  • Add 2-3 composite tools first, not all at once.
  • Each tool combines multiple existing module calls.
  • Partial upstre

Read the thread · 2026-08-19 · closed · 1 comment

Expose MCP resources for commune and dataset reference data

Context

Some data is reference-like rather than action-like: communes, EPCIs, IRIS layers, catalog metadata, and popular datasets. These are good candidates for MCP resources.

Proposed resources

  • reunion://communes
  • reunion://epci
  • reunion://iris
  • reunion://datasets/catalog
  • reunion://datasets/popular

Acceptance criteria

  • Add resource registration using the MCP SDK.
  • Resources return stable JSON with source metadata and timestamps when available.
  • Add tests for resou

Read the thread · 2026-08-19 · closed · 1 comment

Add structured output schemas for every MCP tool

Context

Most tools currently return structured JSON, but the output contract is implicit in implementation code. Adding explicit output documentation/schemas would make the server easier to consume, test, and score on MCP directories.

Scope

  • Inventory every tool output shape.
  • Define stable output contracts for representative repeated patterns: list/search, time series, profile, aggregate, generic catalog.
  • Document returned fields in code comments or generated docs.
  • Add tests that p

Read the thread · 2026-08-19 · closed · 1 comment

[module] Autres (1 datasets)

Tracking issue to wire the Autres datasets from data.regionreunion.com into a dedicated MCP module.

Suggested workflow per dataset

  1. Inspect schema: curl https://data.regionreunion.com/api/explore/v2.1/catalog/datasets/<ID>
  2. Pick the interesting fields.
  3. Add a tool in the corresponding src/modules/*.ts (or create a new module).
  4. Bump TOOL_COUNT in src/modules/index.ts and update README.

Datasets (1)

  • [`frequentation-touristique-mensuelle-a-la-reunion-depuis-20

Read the thread · 2026-04-22 · closed · 0 comments

[module] Sport et loisirs (4 datasets)

Tracking issue to wire the Sport et loisirs datasets from data.regionreunion.com into a dedicated MCP module.

Suggested workflow per dataset

  1. Inspect schema: curl https://data.regionreunion.com/api/explore/v2.1/catalog/datasets/<ID>
  2. Pick the interesting fields.
  3. Add a tool in the corresponding src/modules/*.ts (or create a new module).
  4. Bump TOOL_COUNT in src/modules/index.ts and update README.

Datasets (4)

  • [circuits-rendonnees-lareunion-wssoubik](https:/

Read the thread · 2026-04-22 · closed · 0 comments

[module] Tourisme - Hébergement et restauration (4 datasets)

Tracking issue to wire the Tourisme - Hébergement et restauration datasets from data.regionreunion.com into a dedicated MCP module.

Suggested workflow per dataset

  1. Inspect schema: curl https://data.regionreunion.com/api/explore/v2.1/catalog/datasets/<ID>
  2. Pick the interesting fields.
  3. Add a tool in the corresponding src/modules/*.ts (or create a new module).
  4. Bump TOOL_COUNT in src/modules/index.ts and update README.

Datasets (4)

  • [`etablissements-touristique

Read the thread · 2026-04-22 · closed · 0 comments

[module] Services et social (4 datasets)

Tracking issue to wire the Services et social datasets from data.regionreunion.com into a dedicated MCP module.

Suggested workflow per dataset

  1. Inspect schema: curl https://data.regionreunion.com/api/explore/v2.1/catalog/datasets/<ID>
  2. Pick the interesting fields.
  3. Add a tool in the corresponding src/modules/*.ts (or create a new module).
  4. Bump TOOL_COUNT in src/modules/index.ts and update README.

Datasets (4)

  • [beneficiaires-presta-sociales-caf-lareunion](

Read the thread · 2026-04-22 · closed · 0 comments

The remaining reports are on the project's issue tracker.