dataciviclab
OfficialServer Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| ACB_REPO | No | Sorgente artefatti contesto (default: ACB, branch `context`) | |
| ACB_BRANCH | No | Sorgente artefatti contesto (default: ACB, branch `context`) | |
| GITHUB_TOKEN | No | Discussions (GraphQL richiede auth anche in lettura) + `refresh_context` | |
| DCL_MCP_ENV_FILE | No | Path a un file `.env` (fallback: `.env` cercato verso l'alto dal cwd) |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| session_bootstrapB | Quick orientation — compact markdown overview of Lab state. |
| workspace_triageA | Machine-readable Lab state. Default: compact summary. Use section= to get a specific part: radar, prs, issues, discussions, registry, git, warnings. |
| lab_mapA | Lab navigation map — domains → repos → sources. Use domain= to focus on one editorial theme: an unknown domain returns ok=false with the list of valid domain slugs. |
| findA | Ranked search across datasets, repos, domains, sources, and analyses. All query tokens must match (cross-field AND): words may be spread across slug, name, tags, description, source_id and column schema (names/descriptions/semantic types). Filters: type= (dataset|repo|domain|source|analysis), domain=, limit=. Returns type + parent pointers + score. Use explore(ref) for details. include_github=true (needs GITHUB_TOKEN) adds open issues/PRs whose title contains all query tokens, and top open Discussions of the org on the same query. |
| exploreA | Deep-dive any Lab entity by ref: dataset slug, repo name, domain slug, source id, or analysis slug. Returns card + children + routing hints. |
| github_list_discussionsA | Elenca le GitHub Discussions di un repository (default: hub del Lab). Senza query/category: ordinate per UPDATED_AT discendente, con cursor pagination 'after'/'before'. Con query= e/o category_name: full-text search server-side (mode='search', paginazione 'after'); category_name è un qualifier GitHub, non un filtro lato client. |
| github_list_discussion_categoriesA | Elenca le categorie discussioni valide di un repository (nome, slug, is_answerable) — usala prima di github_create_discussion per non sbagliare category_name. |
| github_get_discussionA | Restituisce una GitHub Discussion. mode=summary restituisce solo un estratto del body (default excerpt_chars=500). mode=full restituisce il body intero (default, backward compat). include_comments=true aggiunge commenti e reply annidate nello stesso round-trip (comments_mode=full|summary, comments_limit, comments_excerpt_chars). |
| github_create_discussionA | Crea una GitHub Discussion in una categoria del repository. Default: repo hub dataciviclab/dataciviclab. category_name accetta nome o slug della categoria (case-insensitive, es. 'Domanda' o 'domanda') — elenca le opzioni con github_list_discussion_categories. |
| github_add_discussion_commentC | Aggiunge un commento top-level a una GitHub Discussion. |
| gcs_queryA | SELECT read-only su un dataset clean o mart pubblicato sui bucket publici del Lab (dataciviclab-clean / dataciviclab-mart). slug= del dataset (layer=clean, default) o mart dataset__table (layer=mart). La query deve referenziare la tabella |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 11 tools
Most tools target clearly distinct actions: the four github_* discussion tools are cleanly separated, and gcs_query vs find vs explore have well-drawn boundaries (query data vs search catalog vs deep-dive a ref). The only soft spot is the trio session_bootstrap / workspace_triage / lab_map, which all return 'state of the Lab' in different formats and could be confused on first use, though the descriptions differentiate them (markdown overview vs machine-readable sections vs domain→repo→source map).
The GitHub subset follows a clean, consistent github_verb_noun pattern (list/get/create/add). The rest mixes bare verbs (find, explore), noun phrases (session_bootstrap, workspace_triage, lab_map), and one verb_noun (gcs_query), so there is no single predictable pattern across the whole set. Names remain readable and snake_case is used uniformly, so it is mixed rather than chaotic.
11 tools is well within a comfortable range and each one earns its place: orientation, catalog search/inspection, one read-only SQL endpoint, and a five-tool discussion workflow. Nothing looks redundant enough to be padding.
The discussion lifecycle is largely covered (list, categories, get with comments/replies, create, comment), and catalog coverage via find/explore/gcs_query is solid. Gaps remain: no update/delete of discussions and no way to reply to a specific comment even though nested replies are returned, and the surface is read-only for issues/PRs and raw/pipeline data (explicitly deferred to another toolkit).