Extractive document summarizer: returns the text's own sentences, so it cannot add facts.
Scored across 1 tool
With only one tool there is nothing to confuse it with, and its description clearly scopes it to extractive summarization rather than generation. An agent can have no doubt about when to select it.
The single name `summarize_document` follows a clear verb_noun snake_case convention and reads exactly as its purpose. There are no competing conventions to clash with.
One tool is thin, but it matches the server's narrow, single-purpose scope (shrink text before passing to a chat model). It earns its place, though the surface is borderline minimal compared to a typical 3-15 tool server.
The core operation is covered, but obvious adjacent operations are absent: no handling of documents beyond the 200k character limit (chunking), no batch/multi-document summarization, and no output-format or ratio-preset options beyond the parameters described. Agents can work around these but will hit the length ceiling.
Tuatara Summarize: extractive summary of a document. Returns the document's own sentences, chosen by a vector-association model, cut to sentence_count, a ratio of the sentences or a token budget; nothing is generated, so the summary cannot add facts. Use it to shrink long text before passing it to a chat model. Up to 200,000 characters per call; no account needed.
| Name | Required | Description | Default |
|---|---|---|---|
| mode | No | auto (default), faithful, gs1, gs2, best, tuned or tuned-best | |
| text | Yes | the document (plain text or HTML) | |
| order | No | document (default: original order) or rank (best first) | |
| ratio | No | share of the sentences to keep, 0.05-0.9 (e.g. 0.2) | |
| max_tokens | No | token budget for the summary (cl100k tokens) | |
| sentence_count | No | number of sentences to keep (1-500) | |
| context_control | No | one word to steer the summary toward a subject | |
| context_control_sensitivity_level | No | 0-10, how strongly the word steers (default 2) |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
The annotations already cover the safety profile (readOnly, idempotent, non-destructive, closed-world), and the description adds the important behavioral facts the annotations don't capture: output is extractive only and cannot introduce facts, the call handles up to 200,000 characters, and no account is required. That's substantive context an agent needs in order to trust the output.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is three tight sentences with no filler. The core identity comes first, then the sizing semantics, then the use case and limits, so the most decision-relevant information is front-loaded.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
With 8 parameters, a 100% schema coverage, and no output schema, the description is complete. It also explains the return values (the document's own sentences, in either original order or ranked order via the schema) and the input-size limit. An agent has everything it needs to call the tool correctly.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, so a baseline of 3 applies. The description goes beyond that by explaining the three mutually exclusive ways to size the output (sentence_count, ratio, or token budget), which the schema lists as separate fields without saying they're alternatives. It doesn't touch mode, order, or context_control, so it isn't a 5.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description gives a specific verb and resource (extractive summary of a document) and immediately distinguishes itself from generative summarizers by explaining the algorithm ('chosen by a vector-association model') and the guarantee ('nothing is generated, so the summary cannot add facts'). No sibling tool competes for this role, so no differentiator is needed.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
It states a concrete use case: 'Use it to shrink long text before passing it to a chat model.' That tells the agent when to reach for it. It doesn't cover exclusions or say what to do for short text, or name a generative alternative when extraction isn't enough, so it falls short of a 5.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
summarize_documentScores text for substance, depth, and clarity in under 25ms.
Strips layout noise via DOM AST; with a query, BM25 filters to relevant sections. No model or API.
Extract clean article text, titles, and metadata from news URLs - ready for NLP pipelines.
Deterministic DOCX/PPTX/XLSX/PDF parser: track changes, comments, headers, footers, merged cells.
Add one secure layer between your agents and this server.
