Marmot Data Catalog

Install

Works with Claude Code, VS Code, Cursor, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install connector/marmotdata-marmot

About

Use cases: database-access, data-analysis, data-pipelines

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this server can do and what it asks for, from its tool list and install config. Checked Oct 25, 2026.

Tools

11 read · 3 write · 0 delete · 1 not yet classified

Secrets

1 (API keys, tokens or passwords)

Install scripts

Nothing to install (hosted)

Runs as

Hosted (remote endpoint)

Tools (15)

Read from the source code; the running server may differ. Descriptions are the server’s own words; hidden characters are shown as [U+…].

• discover_data

  <usecase> Use this to find any data asset in the Marmot catalog. Supports browsing all assets, filtering by type/provider/tags, searching by name and metadata-based queries. </usecase> <instructions> Choose the right approach: - To browse ALL assets: {} (empty parameters) - returns summary breakdown - To count assets: Use filters without query - returns summary with counts - To filter by type: {"types": ["topic", "table", "bucket"]} - returns summary if >20 results - To filter by provider: {"providers": ["kafka", "postgres", "s3"]} - returns summary if >20 results - To search by name: {"query…

• explore_data_products

  <usecase> Use this to explore data products - curated collections of related data assets that together serve a business purpose (e.g. "Customer 360", "Payments Reporting"). Helps answer questions like "what data products exist?", "what assets make up the Customer 360 product?", or "which data products does the payments team own?". </usecase> <instructions> Choose the right approach: - To browse ALL data products: {} (empty parameters) - To search by name/description: {"query": "customer"} - To filter by tags: {"tags": ["gold", "pii"]} - To get a specific product with its assets: {"id": "produ…

• explore_teams

  <usecase> Use this to explore teams and their members. Helps answer questions like "what teams exist?", "who is in the data-engineering team?", or "which teams does john.doe belong to?". For what a team OWNS (assets/terms), use find_ownership instead. </usecase> <instructions> Choose the right approach: - To list ALL teams: {} (empty parameters) - To get a team and its members: {"team_id": "team-uuid"} or {"team_name": "data-engineering"} - To find a user's teams: {"user_id": "user-uuid"} or {"username": "john.doe"} Team details include each member's name, username, email and role. Use offse…

• find_ownership

  <usecase> Use this to answer ownership questions: "who owns this asset?", "what does this user own?", "show me all data owned by the data-eng team". Works bidirectionally - find owners of an asset OR find everything a user/team owns (both data assets and glossary terms). </usecase> <instructions> Provide ONE of: - asset_id: To find who owns a specific asset - user_id or username: To find everything a user owns - team_id or team_name: To find everything a team owns Optional: include_assets (default true), include_glossary_terms (default true) to control what's returned. Returns ownership deta…

• get_asset

  Fetch the full details of a single asset by its Marmot ID. Returns name, MRN, type, provider, description, owner, schema, and provider-specific metadata. Use after search_catalog when you need column/schema details.

• get_upstream_lineage

  Trace the upstream lineage of an asset — what feeds into it. Returns the graph of ancestors up to `depth` hops. Use this to understand where data comes from or to find a root source you can query directly.

• list_queryable_sources

  <usecase> Use this to see which data sources you can run SQL against through Marmot. Call it before query_data to learn the source names. </usecase> <instructions> Optional: {"search": "part of a name or plugin"} to narrow the list. Returns up to 50 sources with each one's name (pass it as "source" to query_data) and the plugin behind it, plus how many match in total; search when there are more than are shown. </instructions>

• lookup_asset

  Look up a single asset by its (type, service, name) triple. Use this when you already know the natural identifiers — for example type='table', service='postgres', name='prod.orders'. Returns null if no asset matches.

• lookup_term

  <usecase> Use this to understand business terminology and definitions. Browse the full glossary, search for terms by name, navigate the term hierarchy, or get a specific term's complete definition including the data assets it's linked to. Helps answer questions like "what does customer_id mean?", "what terms exist under Finance?", or "which assets are tagged with the Revenue term?". </usecase> <instructions> Choose the right approach: - To browse ALL terms: {} (empty parameters) - returns paginated term list - To search by name/definition: {"query": "customer"} - To list children of a term: {…

• manage_owners

  <usecase> Use this to add or remove an owner on an asset. An owner is a user or a team. </usecase> <instructions> Identify the asset with {"id": "asset-id"} or {"mrn": "mrn://..."}, then pass: - "action": "add" or "remove" - "owner_type": "user" or "team" - "owner_id": the id of the user or team (resolve names to ids with explore_teams or find_ownership first) This tool is two-phase and safe by default: - Call WITHOUT "confirm" to PREVIEW the change. Nothing changes. - Only after a human has approved the preview, call again with "confirm": true to apply it. Requires the assets:manage permis…

• manage_tags

  <usecase> Use this to add or remove tags on an asset, for example tagging a table "pii" or "deprecated". </usecase> <instructions> Identify the asset with {"id": "asset-id"} or {"mrn": "mrn://..."}, then pass "add_tags" and/or "remove_tags" as string arrays. Tags already in the desired state are ignored. This tool is two-phase and safe by default: - Call WITHOUT "confirm" to PREVIEW the tags that would be added and removed. Nothing changes. - Only after a human has approved the preview, call again with "confirm": true to apply it. Requires the assets:manage permission. </instructions>

• query_data

  <usecase> Use this to READ data, not just metadata: run one read-only SQL statement against a source through Marmot. The query runs as you, is checked table by table against what you may query, is recorded in Marmot's query log, and returns the rows together with what the catalog says about every table it read (descriptions, tags). </usecase> <instructions> Provide: - source: a name from list_queryable_sources - statement: one read-only SQL statement in the source's own dialect. For Trino, qualify tables as catalog.schema.table, e.g. SELECT * FROM postgresql.public.orders LIMIT 10 - max_row…

• search_catalog

  Search the Marmot data catalog. Returns up to `limit` matches (max 100). `query` accepts plain free text OR Marmot's structured query language. Catalogs can hold millions of assets — prefer structured queries over broad free-text when you know any of: name, type, provider, or metadata. Field filters (combine with AND / OR / NOT, group with parentheses): @type: "Table" - asset type (Table, Topic, Bucket, Alias, Agent...) @provider: "postgres" - source platform (postgres, kafka, OpenSearch...) @name: "users" - exact name match @name contains "customer" - s…

• trace_lineage

  <usecase> Use this to trace data lineage: where data comes from (upstream) and what consumes it (downstream). Helps answer questions like "what feeds this table?", "what breaks if this topic changes?", or "trace the full pipeline around this asset". </usecase> <instructions> Provide ONE of: - asset_id: Trace lineage for an asset by ID - mrn: Trace lineage by MRN (e.g. "postgres://db/schema/table") Optional: - direction: "upstream", "downstream", or "both" (default "both") - depth: How many hops to traverse, 1-10 (default 3) Returns upstream dependencies and downstream consumers grouped by d…

• update_documentation

  <usecase> Use this to write or clear the human-authored documentation on an asset, for example to add a plain-language description of what a table is for. This writes the user_description field and never overwrites the description a plugin ingested, so it is safe for a person's own notes. </usecase> <instructions> Identify the asset with {"id": "asset-id"} or {"mrn": "mrn://..."}, then pass "documentation" with the text to set (an empty string clears it). This tool is two-phase and safe by default: - Call WITHOUT "confirm" (or "confirm": false) to PREVIEW. It returns the current and proposed…

Signals

Stars

623

Forks

32

Open issues

21

Contributors

11

Last commit

Oct 23, 2026 (5 days ago)

Last release

Oct 11, 2026 (17 days ago)

From GitHub (fetched Oct 24, 2026).

Details

Registry name

io.github.marmotdata/marmot

Version

1.0.0

Auth

API key

License

MIT

Repository

github.com/marmotdata/marmot

First seen

Oct 24, 2026

Last checked

Oct 24, 2026

Source

Official MCP Registrygithub-graphql