Skip to content

MCP Tools Reference

The Mention MCP server exposes six tools. This page documents each one — its inputs, what it returns, and when an agent should prefer it. For setup, see MCP Server Overview.

The tools are designed to be used in a flow. Most agents will:

  1. Discover Audiences with list_audiences.
  2. Read the relevant Glossary (read_glossary) or Playbook (list_processes) to find a Concept or Process.
  3. Read the full Article (read_article) or SOP (read_sop) — this is where most of the answer comes from.
  4. Only fall back to ask when the question isn't answered by a single Article or SOP.

list_audiences

Lists all Audiences (knowledge domains) the Agent has access to. This is usually the first call.

Input: none.

Returns: an array of Audience records, each containing:

  • id — the Audience ID, used as audience_id in subsequent calls.
  • name — human-readable name.
  • description — what the Audience covers.

Use this to pick the right Audience for the user's question.

read_glossary

Reads the full Glossary for an Audience.

Input:

  • audience_id (string) — the Audience to read from.

Returns: the Glossary as an ordered list of Concepts. Each Concept includes:

  • id — the Concept ID, used in read_article.
  • name and stored definition.
  • prerequisites — IDs of Concepts that should be understood first.

Use after list_audiences to discover terminology and find the right concept_id. For an initial concept question (what / why / how), prefer read_article next to get the full source content.

read_article

Reads the full Article for a specific Concept. This is the primary Q&A tool for concept questions — Articles contain more content than a synthesized answer can fit and are faster to retrieve once generated.

Input:

  • audience_id (string) — the Audience the Concept belongs to.
  • concept_id (string) — the Concept to read.

Returns: the Article body as structured sections, plus the knowledge-check questions for each section (which highlight key takeaways).

Generation behavior: if the Article has not been generated yet, this call triggers generation and polls until ready. Allow up to roughly one minute for the first read of a new Concept. Subsequent reads return immediately from cache.

list_processes

Lists all Processes in an Audience's Playbook.

Input:

  • audience_id (string) — the Audience to read from.

Returns: an array of Processes, each with:

  • id — the Process ID, used in read_sop.
  • name and short description.

Use to discover which workflows exist and pick the right process_id. For a question about how to perform a Process, prefer read_sop next.

read_sop

Reads the full Standard Operating Procedure for a specific Process. This is the primary Q&A tool for process questions — the SOP contains more content than a synthesized answer can fit.

Input:

  • audience_id (string) — the Audience the Process belongs to.
  • process_id (string) — the Process to read.

Returns: the SOP as a sequence of tasks, with explanatory content, knowledge checks, and any branching workflow.

Generation behavior: if the SOP has not been generated yet, this call triggers generation and polls until ready. SOPs are larger than Articles and can take up to roughly five minutes the first time. Subsequent reads return immediately from cache.

ask

Asks a synthesized question that is answered against the Audience's underlying fact graph and grounded in cited source facts.

Input:

  • audience_id (string) — the Audience to query.
  • question (string) — a natural-language question. More specific questions retrieve more focused answers.

Returns: a synthesized answer with cited source facts.

This is not the primary Q&A tool. Articles and SOPs already contain more material than ask can synthesize, and reading them is faster on already-generated content. Use ask only when:

  • You have already read the relevant Article or SOP and found it insufficient, or
  • The question genuinely spans multiple Concepts or Processes (e.g. "how do X and Y interact?") and no single Article or SOP covers it.

Expect noticeably higher latency than reading already-generated Articles or SOPs.

list_audiences
  └─ pick an audience_id
       │
       ├─ for a concept question:
       │     read_glossary → pick concept_id → read_article
       │
       ├─ for a process question:
       │     list_processes → pick process_id → read_sop
       │
       └─ for a cross-cutting or ambiguous question (after the above):
             ask