tela

Overview Schema Source Related Servers Discussions Admin

Research the wiki

research

Read-only

Answer a question or gather everything on a topic from a team wiki by meaning, returning full page and file excerpts, cited sources, disagreement flags, and a confidence signal.

Instructions

Answer a question — or gather everything relevant on a topic — from the wiki by MEANING. One call assembles answer-ready grounding: the full bodies of the pages that matter (not isolated fragments), pulled from pages AND attached files (PDFs, docs), plus any flagged disagreements among the sources and a low_confidence signal. Returns context (a numbered [n] excerpt block to ground your answer), sources (the cited hits aligned to [n], each with page_id/chunk_id for drill-in and, for file sources, a download_url for the bytes plus a share_url to show the person), disagreements (conflicts to surface, [n]-keyed), low_confidence, and considered/truncated. truncated: true means retrieval found MORE relevant sources than it returned (considered says how many) — the corpus did not run out, the per-call cap did; re-ask with a bigger limit before concluding anything is missing. YOU write the answer from context and cite sources by their [n]. To read one section deeper use read_chunk (chunk_id from a source) or get_page (full page). For exact-name/term lookup use search. Requires a configured embedder (503 otherwise).

Input Schema

Name	Required	Description
`limit`	No	max sources to return (default 12, max 40) — raise it when a result comes back truncated
`question`	Yes	the question to answer, or topic to gather context on
`space_id`	No	optional space id to restrict retrieval to

Output Schema

Name	Required	Description	Default
`context`	Yes
`sources`	Yes
`truncated`	Yes
`considered`	Yes
`disagreements`	No
`low_confidence`	Yes

Schema Changelog

Changes observed during successful MCP inspections.

Changed5 schema fields changedv0.11.0Oct 15, 2026

changedInput schema / properties / limit / description

Previous value: -"optional retrieval depth override (default service-defined)"New value: +"max sources to return (default 12, max 40) — raise it when a result comes back truncated"

addedOutput schema / properties / considered
```
Added value: +{
+  "type": "integer"
+}
```
addedOutput schema / properties / sources / items / properties / share_url
```
Added value: +{
+  "type": "string"
+}
```
addedOutput schema / properties / truncated
```
Added value: +{
+  "type": "boolean"
+}
```

changedOutput schema / required

Previous value: -[
-  "context",
-  "sources",
-  "low_confidence"
-]New value: +[
+  "context",
+  "sources",
+  "low_confidence",
+  "considered",
+  "truncated"
+]

First observedv0.1.0Jul 5, 2026

TDQS

A4.8/5.0

Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations only cover the safety profile (readOnly, non-destructive, closed-world), and the description adds substantial behavioral context beyond them: full bodies rather than fragments, coverage of pages and attached files, surfaced `disagreements` and `low_confidence`, and critically the meaning of `truncated: true` (per-call cap, not exhausted corpus). The embedder/503 dependency is a valuable non-obvious runtime trait.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Front-loads purpose before going into return shape and drill-in routing, and most sentences are functional (truncation semantics, prerequisite, citation by [n]). It is dense and slightly over-enumerates return keys (`context`, `sources`, `disagreements`, `considered`), some of which duplicate the output schema, but the overall organization is purposeful.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

With an output schema present, return-value explanation is optional, yet the description still supplies the semantics an agent needs: how to cite ([n]), how to drill in (chunk_id/download_url/share_url), and how to interpret `truncated`. Combined with the embedder prerequisite and sibling routing, nothing necessary for correct invocation is missing.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so the schema already documents `question`, `space_id` and `limit` (default 12, max 40). The description nonetheless adds actionable semantics for `limit` — raise it before concluding anything is missing when `truncated` is true — which links the parameter to observed behavior rather than just restating its bounds.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb+resource+mechanism: answer/gather on the wiki "by MEANING," with the scope (full page bodies, pages AND attached files) made explicit. It cleanly distinguishes itself from siblings by naming `search` as exact-term lookup and positioning itself as the semantic alternative.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Explicit routing: use this when you need answer-ready grounding, `read_chunk` to read one section deeper, `get_page` for a full page, `search` for exact-name/term lookup. It also states the operational prerequisite — a configured embedder is required (503 otherwise) — and tells the agent how to react to truncation.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.