Resources · Claude connector

Claude Within connectorWhat Claude can do in your Within account.

The Claude Within connector exposes 17 tools: 17 read tools, 0 write tools. They walk your organization's process index, field observations and source documents. Below: what Claude can find, what stays in Within, and who has to switch access on.

Verified Trustpilot reviews · AI, automation & growth agency

Overview

What changes when you need to know how work really gets done

Within maps how work actually happens in your company: a process index (hierarchy, steps, dependencies) plus the documents behind it. Connected to Claude, that index answers plain-language questions with evidence pulled from your own workspace instead of generic guesses.

Land on the right process. search runs a semantic search over the index, then fetch pulls one process in full: its place in the index, its current version with steps, policies and linked documents, its dependencies and its history.

See what happens on the ground. get_process_observations surfaces captured runs of a process with their friction signals, and get_observation_citation replays the activity timeline behind one of them: who did what, when, on which systems.

Quote the exact source. search_artifacts finds a document, and get_artifact_content extracts its text, by line range when you need to cite a precise passage.

What Claude does not do here, per Within: create or alter a process, a document or a graph entity. That happens in the Within interface, along with bulk edits, workspace configuration and user management. And nothing runs on its own: no tool reacts when a process evolves. That kind of watching is a job for an automation tool, a different trade; the Integrations hub covers it, and the Claude connectors hub lists the others.

Vocabulary

The vocabulary in one minute

Five words you will meet while connecting Within.

Connector
The link you set up once between Claude and an account you already own, so Claude can work in it while it answers you.
Tool
One named thing a connector lets Claude do. Claude picks the ones it needs on its own; the directory lists them by name.
Authorization
The service's own sign-in screen, where you hand Claude the access it will use. Granted once per person, and you can take it back.
Approval
The confirmation Claude waits for before it finishes something that changes your account, shown in the chat when it matters.
MCP
The shared standard connectors are built on: it is what lets an assistant talk to an outside service. You will meet the acronym elsewhere.
Connect

Connect Within to Claude in three steps

  1. 01

    Find Within in Claude

    In Claude's settings, open Customize, then Connectors, and search for Within. On a Team or Enterprise workspace, an Owner or Primary Owner first has to enable the connector for the organization.

  2. 02

    Start the connection

    Click Connect on its row: Within opens its own sign-in page, where you authenticate. Claude then reaches your active workspace. If the link breaks later, use Disconnect and connect again.

  3. 03

    Read the authorization screen

    Read the authorization screen before you accept: it belongs to Within, not to Claude, and it sets what the access covers. It may list broader permissions than what Claude does day to day; per-action approval stays the default rule.

Tools

The 17 tools, grouped by what they explore

Within gives Claude 17 tools: 17 that read your account, 0 that change something in it.

Seventeen tools, all described by Within as lookups. Names stay exactly as Claude shows them.

  • 17 read

What Claude reads (17)

17 tools

Process index, observations, documents, workspaces and Agent Builder: every tool here is a lookup.

fetch

Pulls one process in full from an ID that search or the hierarchy returned: where it sits in the index, its current version with steps, policies and linked documents, its dependencies and its history.

When it helps
a search has surfaced the process you were after and you want the whole record behind it.

Sourcedevelopers.within.ai · October 1, 2026 ↗

get_agent_builder_instructions

Fetches the instructions that go with Agent Builder, which Within describes as a guided way to turn how your organization works into skills, agents and automations.

When it helps
you are working through Agent Builder and Claude needs its instructions before going further.

Sourcedevelopers.within.ai · October 1, 2026 ↗

get_agent_builder_resource

Retrieves one resource used while building with Agent Builder. Within's documentation stops at that line and does not say what those resources contain.

When it helps
an Agent Builder build is under way and Claude needs one of the resources it relies on.
Watch out
no source describes this tool beyond that single line, so expect nothing more than it states.

Sourcedevelopers.within.ai · October 1, 2026 ↗

get_artifact_content

Extracts the text of one document, in full or by line range. Within points to range mode for the moments when specific lines have to be cited.

When it helps
you need the exact wording of a passage from one document rather than a summary of it.

Sourcedevelopers.within.ai · October 1, 2026 ↗

get_artifact_details

Returns the full metadata of one document identified by its key: its type, its status, its file information and the documents linked to it.

When it helps
you want a document's status and its links before deciding whether to read it in full.

Sourcedevelopers.within.ai · October 1, 2026 ↗

get_hierarchy_node_details

Inspects a single node of the index and shows its parent, its children, the process attached to it and its attributes.

When it helps
you want to see what sits under one part of the index before going a level further down.

Sourcedevelopers.within.ai · October 1, 2026 ↗

get_observation_citation

Replays the activity timeline behind an observation: what the person did, when, and on which systems. Within calls it the most primary-source evidence it holds.

When it helps
an observation raises a question and you want to see the activity that produced it.
Watch out
it only works for your own sessions.

Sourcedevelopers.within.ai · October 1, 2026 ↗

get_process_details

Returns the rich nested record of a process: its current version, its dependencies, its hierarchy node, its attributes and, as an option, its version history.

When it helps
you need a process's dependencies and attributes gathered in one structured answer.

Sourcedevelopers.within.ai · October 1, 2026 ↗

get_process_hierarchy

Walks the whole structure of the process index. Within suggests it when search misses, or to get your bearings before drilling into one process.

When it helps
search turned up nothing convincing and you would rather find your way through the tree of processes.

Sourcedevelopers.within.ai · October 1, 2026 ↗

get_process_observations

Lists the observations recorded on one process. Each carries its friction signals (rationale, time frame, negative and positive deviations) and a reference to the session it came from.

When it helps
you want to see where a process departs, in practice, from the version written down in the index.

Sourcedevelopers.within.ai · October 1, 2026 ↗

get_recent_process_changes

Gives the feed of new process versions across the whole workspace. Within frames it around a single question: what has been edited recently?

When it helps
you want to know which processes were edited lately, anywhere in the workspace, without opening each of them one by one.

Sourcedevelopers.within.ai · October 1, 2026 ↗

list_accessible_workspaces

Lists the Within workspaces open to you and shows which one is currently active. Each connection is scoped to one of them, according to Within.

When it helps
you belong to more than one workspace and want to confirm which one Claude is reading before you dig in.

Sourcedevelopers.within.ai · October 1, 2026 ↗

list_sessions

Lists recent sessions in the workspace, page by page. Each entry shows the processes the session updated, its duration and the person who created it.

When it helps
you want a view of recent activity in the workspace and of the processes it touched.

Sourcedevelopers.within.ai · October 1, 2026 ↗

search_artifact_text

Searches inside a single document, identified by its key, with a regular expression or literal text, and returns line-numbered matches with their surrounding context.

When it helps
you need every line of one long document that mentions a given term.
Watch out
Within states there is no cross-document text search on this side.

Sourcedevelopers.within.ai · October 1, 2026 ↗

search_artifacts

Runs a hybrid search, by meaning and by words, across the text of documents, and returns ranked results with a short context snippet for each.

When it helps
you are looking for a document without knowing where it is filed in the workspace.

Sourcedevelopers.within.ai · October 1, 2026 ↗

search_processes

Ranks processes by combining keyword matches on name, objective, team and value stream with semantic similarity.

When it helps
you know the team or value stream a process belongs to, but not the exact name it carries in the index.

Sourcedevelopers.within.ai · October 1, 2026 ↗

Approvals

What Claude asks before it acts

By default, Claude stops and asks before every action it takes on an account on your behalf. Within describes all seventeen tools as lookups, and no source describes a confirmation specific to any of them.

On Team and Enterprise, workspace owners decide whether a member may let some actions through without approving each one. They can also cap what a connector may do for the whole organization, and no one gets around that from a personal account. On the Within side, each connection is scoped to a single workspace, your active one when you connect. The official read versus write label for each tool only shows up in the settings of a connected account.

Plans

Which plans include it

None of the 819 sheets in the official directory shows plan availability. Connector by connector, that information is not published anywhere.

The general rule is published: remote connectors are open to all users on Claude, Cowork, Claude Desktop and mobile. On Team and Enterprise, an Owner or Primary Owner enables them for the organization first. On the Within side, you need an active workspace and a user account. For the connector's current status on your account, check its sheet in the official directory.

Limits

Where this connector stops

A connector is not an automation. Claude calls these tools while it answers you; nothing fires when a process evolves in Within.

Only the publisher documents this connector; Anthropic has no page about it. By default, its sheet only exposes read tools, but that is an observed floor, not a guarantee. The directory badge is not a security audit either, and Anthropic writes on every sheet that it does not choose the tools a publisher exposes. Within itself warns that its catalog keeps growing, including write tools for some operations it considers safe.

Need help

Need help connecting Within to Claude?

A person reads every message.

FAQ

FAQ about the Claude Within connector

01What can Claude do with the Within connector?
Claude can query your organization's process index and the documents attached to it. Its seventeen tools cover process search, hierarchy navigation, versions and dependencies, captured observations and their activity timelines, document search and reading, workspace and session listings, plus two tools for the Agent Builder flow. Within describes every one of them as a lookup, so the connector answers questions rather than editing your processes, and each answer can point back to the record it came from.
02Can Claude edit my processes in Within?
No, not according to Within: its production tools are lookups, and creating or altering processes, documents or graph entities happens in the Within interface. The publisher adds that its catalog grows over time, eventually including write tools for some safe operations. So today's read-only list is an observed floor, not a promise about tomorrow; the directory sheet shows the current state. Within also points to its interface for bulk edits, workspace settings and user management.
03Does Claude ask before it acts with Within?
By default, Claude asks for your go-ahead before every action it takes on an account for you. For Within, no source describes a confirmation tied to any of the seventeen tools, all presented as lookups by the publisher. On a Team or Enterprise workspace, owners decide whether some actions may pass without a prompt, and they can restrict the connector for everyone in the organization, which no member can undo alone from their own account.
04Which plans is it available on?
No official source publishes plan availability connector by connector, and none of the 819 directory sheets displays it. The general rule says remote connectors are open to all users on Claude, Cowork, Claude Desktop and mobile. On Team and Enterprise, an Owner or Primary Owner first has to enable the connector for the organization. On the Within side, you also need an active workspace and a user account, since Within is a commercial service for its customers.
05Does Claude see every Within workspace in my company?
No. Each connection is scoped to a single Within workspace, the one that is active at the moment you connect. Claude does not jump between workspaces on its own. If you belong to several, Within says you switch by disconnecting and reconnecting. The list_accessible_workspaces tool lets Claude tell you which one is active. The detailed timeline behind an observation is only available for your own sessions, not for your colleagues' sessions.
06Why can't Claude find a process I know exists?
Because search runs by meaning over your process index, and a single query can miss. Within sets out what to try next: when search misses, get_process_hierarchy browses the whole structure of the index so Claude can find its bearings. search_processes ranks processes by mixing keyword matches on name, objective, team and value stream with semantic similarity. And when what you want is a document rather than a process, search_artifacts runs a hybrid search across document text.
07Claude or an automation tool for Within?
They do different jobs, so choose by use. Claude fits when you have a question about a process and want an answer backed by your own data: how a task is done, what moved recently, where things stall. An automation tool fits when something must fire on its own, without a question, on every new version for instance. Claude only acts during the conversation and watches nothing in between, so the two cover different needs.