Skip to main content
Glama

Server Details

Cited answers in Claude or ChatGPT only from your office's procedures; says so when not covered.

Ownership verified
Status
Healthy
Uptime
100.0% over the last 24 hours
OAuth
Works in Glama
Last Tested
Transport
Streamable HTTP · MCP 2025-11-25
URL
Repository
forgeagents-jp/oka-anywhere
GitHub Stars
0
ChatGPT
–
Claude
OKA Anywhere

TDQS

A4.5/5.0

Scored across 4 tools

Disambiguation5/5

Each tool has a clearly distinct role: LLM-synthesized answer, keyword-based passage retrieval, verbatim quote verification, and enumeration of documented boundaries/prohibitions. The descriptions explicitly contrast the overlapping-looking pair (ask vs. search) by calling out 'uses a language model' vs. 'without a language model' and by describing different return shapes. An agent should be able to select correctly.

Naming Consistency5/5

All four names follow a verb_noun pattern in snake_case with a consistent 'office_procedure(s)' object noun: ask_/list_/search_/verify_. The slight variation between 'policies_and_procedures' and 'procedure_boundaries'/'procedure_quote' is natural phrasing, not a convention break.

Tool Count5/5

Four tightly scoped, read-only tools for a document-grounded Q&A server, and each one covers a genuinely different mode (synthesize, retrieve, verify, enumerate gaps). Nothing appears redundant or missing at the count level.

Completeness4/5

The core lifecycle for grounded document Q&A is covered: answer, find passages, verify exact quotes, and discover documented gaps/limits. The main gap is corpus navigation — there is no tool to list available documents or fetch a full document by title, so an agent can only see at most three excerpts per search and cannot browse the document set.

Publisher details

Operator
Forge Agents LLC · Publisher source
Vendor relationship
First-party
Privacy policy
Not provided
Trust center
Not applicable
Restrictions
Paid plan: $149 a month per office, with a 14-day free trial through Stripe checkout at forgeagentsio.com/oka.html. Each office connects with its own office token, which it gets on the welcome page after checkout; each person who connects pastes it once. Built for Claude and ChatGPT; other MCP clients may work but are not promised. · Publisher source

Available Tools

4 tools
ask_office_policies_and_proceduresAsk the office proceduresA
Read-onlyIdempotent
Inspect

Answer a question about how this office does something, in plain language, using only the office's own approved procedure documents. Use it when the user asks how their office handles something those documents may cover: a policy, a step, a fee, a deadline, a form, who approves or who to ask. The user does not need to name the documents. Not for questions that have nothing to do with the office's own procedures, such as general knowledge, writing or math. Returns a written answer followed by a Sources list of verbatim quotes that were checked against the documents. If the documents do not cover the question it says so instead of guessing, and rephrasing will not change that. Uses a language model to write the answer.

Inputs: question (required), corpus.

list_office_procedure_boundariesList what is not coveredA
Read-onlyIdempotent
Inspect

List what the office's approved documentation explicitly forbids or says it does not cover. Use it before concluding that something is allowed, or when a question is not covered, to see whether the topic is a documented gap. Returns the boundary statements grouped by document title, or a note that none were found. Takes no question and uses no language model.

Inputs: corpus.

search_office_policies_and_proceduresSearch the proceduresA
Read-onlyIdempotent
Inspect

Find the most relevant passages in the office's approved procedure documents for a topic, without a language model. Use it to read the source text, to see what exists on a topic, or when an exact sentence is not found. Returns up to three document titles with an excerpt each, or says no matching procedure was found. Ranking is by keyword overlap, so use the words the documents are likely to use.

Inputs: query (required), corpus.

verify_office_procedure_quoteVerify an exact claimA
Read-onlyIdempotent
Inspect

Check whether an exact statement appears word for word in the office's approved procedure documents. Use it to confirm a specific sentence, number or rule before relying on it. Returns VERIFIED with the document title and the surrounding quote, or says the statement was not found verbatim. VERIFIED means the whole statement was found together in one document; when the sentences of a long statement are found only in part, or only apart, it says not found and shows the partial match. Case, spacing and curly quotes are ignored, but a paraphrase will not match. Uses no language model.

Inputs: claim (required), corpus.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • F
      license
      Not graded
      quality
      C
      maintenance
      Enables retrieval-augmented question answering over company policy documents via MCP tools and resources, with optional Claude-generated grounded summaries.
      -
    • F
      license
      Not graded
      quality
      C
      maintenance
      Enables AI agents to query a governed knowledge base and receive cited answers, or explicitly decline when the record does not cover the question, preventing improvisation over gaps.
      -
    • A
      license
      Not graded
      quality
      C
      maintenance
      Enables Claude to query client and document data in plain English with cited answers, enforce per-client access, safely file incoming documents, and record audited actions through MCP tools.
      MIT
    Try in Browser

    Glama MCP Gateway

    Add one secure layer between your agents and this server.