# Use the governed definition of a metric or term before answering a data question

> Use the governed definition of a metric or term before answering a data question: 0 companies fully do it, 9 with a limit, 34 carry the tag. Every listing in Company knowledge & data catalogues with the reason and a source.

- Canonical: https://www.anchorterminal.com/niches/company-knowledge.metric-definition/
- Markdown: https://www.anchorterminal.com/niches/company-knowledge.metric-definition/index.md (~1,650 tokens)
- Slim: https://www.anchorterminal.com/niches/company-knowledge.metric-definition/index.min.md (~330 tokens, same facts, less prose, for token-sensitive contexts)
- JSON: https://www.anchorterminal.com/niches/company-knowledge.metric-definition/index.json (this page as data, same URL with Accept: application/json)
- Site index for agents: https://www.anchorterminal.com/llms.txt (the key docs and every listing in one file, at most 16 MB, about 4 million tokens: https://www.anchorterminal.com/llms-full.txt)
- API: https://www.anchorterminal.com/api/v1/index.json
- MCP: https://www.anchorterminal.com/mcp (the whole catalogue as tools, no key)
- DNS: dig +short TXT <slug>.grade.anchorterminal.com (the grade as a DNS TXT record, for a sandbox with no HTTP where names still resolve; also top., best.<category>. and <a>.vs.<b>. under the zone, and the same under grade.letme.dev)
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)
- Updated: 2026-11-15

No company's docs show a full fit yet. The listings below do the job with a limit, or not at all.

| Companies, full fit | With a limit | Carry the tag | Share of the category's work | First-hand accounts |
| --- | --- | --- | --- | --- |
| 0 | 9 | 34 | 7 | 3 |

## What counts as done

The glossary or metric definition with its owner and the query or model that implements it, then a read-only query answer citing that definition; ad hoc SQL capped by a row limit.

Measured, for our own tests: done within three calls.

How agents ask for it: what counts as an active customer here; answer churn last quarter using our definition; what does ARR mean in finance's glossary

## Does the job, with a limit

It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.

- **Alation (Alation, Inc.)**: sql_query_agent and query_flow_agent generate SQL against a data product with analysis and execution guidance, but the pages read show no statement guard or row cap, and calls draw on the 2,000-call quota.
- **Atlan (Atlan Pte. Ltd.)**: Glossary terms come back with assets, and query_assets runs read-only SQL capped at 100 rows under the caller's query permission, but no tool links a term to the query that implements it, and the open-source toolkit has an open read-only enforcement bug (issue 226).
- **Basedash MCP (BaseDash Inc. (Delaware, United States))**: Models define measures with a verified flag and the AI queries them read-only, returning the answer with CSV; but whether the MCP reply names the model owner and SQL is unstated, no row cap is given, and each question spends AI credits at an unpublished rate.
- **ClicData MCP (ClicData SAS)**: The page says answers trace back to the governed warehouse, but no glossary or metric-definition tool, owner field, read-only rule or row cap is documented in what could be read, and access is early access through a demo request.
- **Collibra (Collibra Inc.)**: discover_business_glossary returns definitions and get_measure_data traces a measure to its columns, but no tool runs a query, so the read-only answer citing the definition needs another tool.
- **DataHub (Acryl Data, Inc. (d/b/a DataHub))**: Glossary terms, term versions and real queries per dataset come back, and draft_sql_for_tables drafts SQL, but no tool runs a query, so the read-only answer citing the definition needs another tool.
- **Marmot (Marmot Data Ltd)**: lookup_term gives a term's definition and linked assets, and query_data runs one read-only statement as the caller with a row cap (default 1,000, cap 10,000), but the term's implementing query isn't returned and its owner needs find_ownership.
- **OpenMetadata (Collate, Inc.)**: find_context then get_concept_context return a term or metric's definition, column bindings and saved-query evidence in two calls, but no tool runs SQL, so the read-only answer needs another tool.
- **Secoda (Secoda, Inc.)**: glossary returns the whole glossary and search_documentation finds terms, then run_sql answers, but the repo applies no read-only check or row cap; the prompt only suggests LIMIT, and owner and implementing query fields are unshown.

## Couldn't check

The docs were closed to us, or silent. These carry no mark either way.

- **Precisely AI Studio (Precisely Incorporated)**: No glossary or metric-definition asset is named on the readable page; the help centre couldn't be read.

- not this job: Cohere North (Cohere Inc.): No glossary or governed metric layer described; the financial data providers named are planned connectors, not definitions.
- not this job: Dust (Permutation Labs): No glossary or governed metric layer; definitions could only be found as document text.
- not this job: Gemini Enterprise (Google LLC): No governed glossary or metric definitions with owners and implementing queries in the StreamAssist or connector docs.
- not this job: Glean (Glean Technologies, Inc.): No governed glossary or metric definitions with owners. data_analysis routes analytical questions to a BI tool, but returns no definition and implementing query first.
- not this job: Guru (Guru Technologies, Inc.): A verified card can hold a definition, but there is no glossary object tied to a query and no SQL, so the read-only query step can't be met.
- not this job: Hexis Cloud (Bevelites GmbH): A page can hold a definition, but there is no glossary object linked to a query and no SQL tool.
- not this job: Onyx (DanswerAI, Inc.): No glossary, semantic layer or SQL tool on the MCP server; a definition could only be found as ordinary document text.
- not this job: Overclock (Overclock Technologies Limited): No glossary, metric layer or SQL; facts are extracted from documents.
- not this job: Procore AI Go (Procore Technologies, Inc.): No glossary, semantic layer or SQL; the app answers project-record questions by voice.
- not this job: Lens (Thought Industries) (Thought Industries, Inc.): No glossary, metric layer or SQL; the API asks questions and manages sources.

## What stops agents finishing

- each question spends AI credits with no published per-question price (Basedash)
- free API quota that never resets (Alation, 2,000 calls)
- MCP early access only through a demo request (ClicData)

## Around it

Before: chat.user-question. After: analytics.chart, databases.sql-query.

## Where this comes from

Weight 7 of 100 in Company knowledge & data catalogues: 3 of 38 counted accounts (dbt definitions pushed for agents, internal data-lake assistant, Atlan glossary creation); Basedash, ClicData, Atlan and Alation pages make this their main claim, so vendor pages carry more of the weight here

As data: https://www.anchorterminal.com/api/v1/use-cases/company-knowledge.metric-definition.json, or the MCP tool get_use_case with id company-knowledge.metric-definition. Draft, read 2026-11-15 from public docs and first-hand accounts, not yet reviewed by a person.

---

Query this catalogue directly over MCP: https://www.anchorterminal.com/mcp (Streamable HTTP, no key). How: https://www.anchorterminal.com/agents/#mcp
