4.8
on G2
Install Now
Add Docsie's MCP server to Claude Desktop or Claude Code with one config block.
Or paste your Docsie workspace URL
Works with Claude Desktop, Claude Code, and Claude API clients
OAuth 2.0, RBAC, audit logging, and on-prem/private deployment options.
Claude was built around MCP — Anthropic's open standard. Docsie's MCP server connects Claude Desktop and Claude Code to private docs. Add a config block to claude_desktop_config.json and Claude can ground answers in your internal documentation with permission-aware access and reviewable logs.
Engineering, support, and platform teams use Docsie's MCP server to ground Claude Desktop and Claude Code in real internal documentation
_Leader_Small-Business_Leader.svg)
_BestSupport_Small-Business_QualityOfSupport.svg)
Claude MCP Options Compared
Claude supports many MCP servers. Here's how Docsie compares for grounding Claude in your private enterprise docs.
| Claude MCP Integration Feature |
Docsie MCP
Private Docs
|
Context7
|
Filesystem MCP
|
Anthropic Files API
|
DIY RAG MCP
|
|---|---|---|---|---|---|
| Native MCP server (Anthropic spec) | |||||
| Private enterprise documentation | |||||
| Works with Claude Desktop | |||||
| Works with Claude Code | |||||
| OAuth 2.0 + enterprise SSO | |||||
| RBAC inherited from doc permissions | |||||
| Real-time doc sync (no re-indexing) | |||||
| Audit log of Claude MCP queries | |||||
| Setup effort | Standard Claude config | Standard MCP config | Local file setup | Custom build | Custom build |
| Enterprise security and privacy controls |
Comparison based on publicly documented Claude MCP options as of June 2026.
Claude + Docsie MCP Impact
Here's what changes when Claude has direct access to your internal documentation via Docsie MCP vs. answering from training data.
How to Add Docsie MCP to Claude
Add a config block to Claude Desktop or Claude Code, sign in, and let Claude query governed documentation context.
Open claude_desktop_config.json (Claude Desktop) or your Claude Code MCP config. Add the Docsie MCP server block with your workspace URL. Save and restart Claude — the new MCP server is now available.
First Claude conversation that triggers an MCP query opens an OAuth window. Sign in through your enterprise SSO (Azure AD, Okta, Google, SAML). Claude's MCP token is per-user — scoped to your individual permissions.
From now on, Claude questions that need internal context can query Docsie via MCP. Claude can cite source docs, use internal specifics, and ground answers in approved documentation instead of relying only on training data.
Why Claude Teams Use Docsie MCP
Claude is Anthropic's flagship — and MCP is Anthropic's open standard. Docsie's MCP server is the native private-docs companion built for both.
Docsie is built around the MCP pattern for making documentation available to AI tools. Claude Desktop, Claude Code, and MCP-compatible Anthropic workflows can connect through standard MCP configuration.
Ask Claude about internal systems, APIs, processes, or policies. Claude can call docsie.search via MCP, retrieve internal docs, and answer with source context.
Each user's Claude inherits their Docsie permissions. Senior engineers see senior-only architecture docs; junior engineers don't. HR managers see HR docs; everyone else doesn't. RBAC is enforced server-side.
Update an internal SOP and configured MCP workflows can make the newer version available to Claude without a separate embedding-refresh project.
Claude docsie.search and docsie.fetch calls can be logged. Compliance teams can see who queried what, when, and which docs were returned, supporting SOC 2, ISO 27001, and GDPR review evidence.
Add a config file, restart Claude, and sign in via OAuth. Teams can avoid building a separate embeddings pipeline or vector database for this documentation workflow.
Engineering, support, and platform teams use Docsie's MCP server to make Claude genuinely useful for their internal documentation
Claude Code is Anthropic's AI coding agent — and it shines when grounded in real internal APIs. Connect Claude Code to Docsie MCP, and every code generation prompt auto-queries your internal API references, ADRs, and runbooks for real internal context.
Support teams using Claude in their workflow can ground customer answers in the real KB. Claude can query Docsie MCP for help center articles, cite the source, and help agents stay aligned with policy.
Internal teams use Claude as a personal AI assistant. Connect it to Docsie MCP and Claude can answer PTO, benefits, expense policy, and IT process questions from your real internal docs — RBAC-scoped to what each employee should see.
Common Questions
Everything you need to know about connecting Claude Desktop and Claude Code to Docsie via the Model Context Protocol
Q: How do I add the Docsie MCP server to Claude Desktop?
A: Open Claude Desktop's config file (claude_desktop_config.json — location varies by OS, accessible from Claude's Settings → Developer menu). Add the Docsie MCP server block with your workspace URL. Save and restart Claude Desktop. The first Claude conversation that triggers an MCP query will open an OAuth sign-in window through your enterprise SSO. After sign-in, Claude uses Docsie MCP automatically for any prompt that needs internal context. Setup is typically a short flow.
Q: Does this work with Claude Code (the CLI)?
A: Yes. Claude Code is Anthropic's AI coding agent and supports MCP servers natively. Add the Docsie MCP server to your Claude Code config (~/.claude/config.json or per-project config). Same OAuth flow, same RBAC, same audit log. Claude Code uses Docsie MCP for any prompt that touches internal services, APIs, or docs.
Q: Why is Claude particularly good with MCP?
A: Anthropic created the Model Context Protocol as an open standard for connecting AI models to external context. Claude has strong MCP support, and Docsie implements the MCP pattern for documentation access.
Q: Can I use Docsie MCP with the Claude API directly?
A: Applications built on Claude can use MCP-style tool sources where the client architecture supports MCP. Docsie's MCP server exposes docsie.search and docsie.fetch through the standard protocol.
Q: How does Claude authenticate to Docsie MCP?
A: OAuth 2.0. First MCP query triggers a browser-based sign-in through your enterprise SSO (Azure AD, Okta, Google, SAML). Docsie issues a per-user access token scoped to your identity. The token is stored securely by Claude (in its MCP credential store) and used for subsequent queries. Tokens refresh automatically; you can revoke at any time from Docsie's admin dashboard or via SSO deprovisioning.
Q: Are Claude's MCP queries audit-logged?
A: Yes. MCP queries Claude makes can be logged with user identity, timestamp, query parameters, and documents returned. Logs can support SOC 2, ISO 27001, GDPR, and HIPAA-sensitive reviews of AI agent usage and can be exported to SIEM tools such as Splunk, Datadog, or Elastic.
Q: Can Claude see docs the user shouldn't?
A: No. Docsie's MCP server enforces RBAC at the API layer. When Claude calls docsie.search via MCP, the server runs the query scoped to the authenticated user's permissions. If the user cannot see a doc in Docsie, Claude responses are scoped away from that content. There's no agent-side bypass possible — the permission filter is enforced server-side.
Q: What kinds of work are best for Claude + Docsie MCP?
A: Best fits: technical Q&A about internal systems, code generation grounded in internal APIs, drafting docs that reference internal policies, summarizing internal procedures, answering compliance questions from real internal SOPs, customer support grounded in real KB articles. Anything where Claude needs real internal context to give a grounded answer, not a generic one.
Q: How does this compare to Claude Projects with file uploads?
A: Claude Projects let you upload specific files for a project context. Docsie MCP gives Claude governed access to your documentation library, with permissions and audit. Claude Projects work well for ad-hoc one-off contexts; Docsie MCP is the right pattern for grounding Claude in living internal docs that you cannot manually upload to every project.
Q: Can we standardize Docsie MCP across our whole Claude install base?
A: Yes. For Claude Desktop, IT teams can push a standard claude_desktop_config.json via MDM (Jamf, Intune, Kandji) so every employee's Claude has the Docsie MCP server pre-configured. For Claude Code, per-repo config in .claude/config.json means every developer working in your codebase gets the same MCP setup.
Ready to ground Claude in your real internal docs?
Book a DemoDrop one config block into claude_desktop_config.json. Give Claude Desktop and Claude Code audit-logged access to your private internal docs through configured OAuth and RBAC.
Works with Claude Desktop, Claude Code, and Claude API clients. OAuth 2.0 authentication.