Phabricator MCP. Manage dev tasks, code reviews, and repos in chat.
Gemini Works with every AI agent you already use
…and any MCP-compatible client
Just plug in your AI agents and start using Vinkius.
Phabricator (Development Platform Conduit API) connects your AI agent directly to Phabricator's core services. You can search Maniphest tasks using complex filters, edit code revisions in Differential, and query repository data from Diffusion—all without leaving your IDE or chat client.
What your AI agents can do
Differential query
Queries historical code revisions stored in Differential (Legacy).
Differential revision edit
Creates or modifies a specific code revision record within Differential.
Differential revision search
Searches for relevant code revisions in Differential using filters.
Find specific Maniphest tasks based on complex criteria, create new tasks, update existing ones (status/owner), or retrieve a full history of any task's lifecycle.
Query for code diff revisions in Differential and perform direct actions like requesting changes to the review.
Search across Diffusion repositories to find specific commits or list available project repositories.
Get details about the currently logged-in user or search for other people associated with tasks and code reviews.
Ask AI about this MCP
Supported MCP Clients
Gemini Waiting for input…
Phabricator (Development Platform Conduit API): 10 Tools
These tools let your AI agent search tasks, edit revisions, query commits, and manage user data directly from Phabricator.
019e5d44differential query
Queries historical code revisions stored in Differential (Legacy).
019e5d44differential revision edit
Creates or modifies a specific code revision record within Differential.
019e5d44differential revision search
Searches for relevant code revisions in Differential using filters.
019e5d44diffusion commit search
Searches for specific commit IDs across Diffusion repositories.
019e5d44diffusion repository search
Lists and searches available code repositories within Diffusion.
019e5d44maniphest edit
Creates a new Maniphest task or updates an existing one (e.g., changing status or assigning an owner).
019e5d44maniphest gettasktransactions
Retrieves the complete, chronological activity history for any given task.
019e5d44maniphest search
Searches Maniphest tasks using advanced filtering rules and constraints.
019e5d44user search
Finds details about other users who contribute to the platform.
019e5d44user whoami
Gets basic profile information for the user who is currently logged into Phabricator.
Choose How to Get Started
Build a custom MCP for your own tools, or connect a ready-made integration from our catalog.
Build Your Own
Turn any API into an MCP. Import a spec, define Agent Skills, or deploy with MCPFusion.
- Import from OpenAPI, Swagger, or YAML specs
- Create Agent Skills with progressive disclosure
- Deploy to edge with MCPFusion framework
- Built in DLP, auth, and compliance on every call
- Real time usage dashboard and cost metering
- Publish to catalog or keep private
Make Your AI Do More
Start with Phabricator (Development Platform Conduit API), then connect any of our 4,700+ other servers whenever your AI needs more. One click, no limits.
- Use this MCP plus 4,700+ others, all in one place
- Add new capabilities to your AI anytime you want
- Every connection is secured and compliant automatically
- Track usage and costs across all your servers
- Works with Claude, ChatGPT, Cursor, and more
- New servers added to the catalog every week
What you can do with this MCP connector
Phabricator Conduit API connects your AI agent straight into Phabricator's core services. You get direct access to Maniphest for managing tasks, Differential for handling code reviews, and Diffusion for inspecting repository history—all without ever leaving your IDE or chat client.
Managing Tasks and Auditing Workflows
You can locate specific tasks by running maniphest_search with advanced filtering rules. If you need to see the full life cycle of a task, maniphest_gettasktransactions pulls every single piece of activity history into one chronological record. You'll use user_whoami to check your own profile details, and if you gotta look up another contributor, user_search finds their details on the platform.
When it comes time to move things along or fix something, you can either create a brand-new Maniphest task or update an existing one using maniphest_edit; this lets your agent change statuses or assign owners. It's basically complete control over the task lifecycle right from your chat client.
Handling Code Reviews in Differential
Working with code revisions is simple. To look at old diff revisions, you run differential_revision_search to find relevant records. If you spot an error or a change that needs adjustment during the review process, differential_revision_edit lets your agent modify that specific code revision record directly.
Inspecting Repository History with Diffusion
Diffusion gives you deep visibility into your code base. You first run diffusion_repository_search to list all available projects and repositories. Once you know which project you're dealing with, you can use diffusion_commit_search to track down specific commit IDs across those repositories.
How Phabricator MCP Works
- 1 First, subscribe to this server on Vinkius. You'll need your Phabricator Instance URL and a Conduit API Token.
- 2 Next, configure your AI client (Claude, Cursor, etc.) with those credentials. The connection establishes access to all exposed tools.
- 3 Finally, ask your agent to perform an action—like 'Find me the history for task T101'—and it uses the appropriate tool endpoint.
The bottom line is: you tell your AI client what dev data you need; it calls the API and brings back the result instantly.
Who Is Phabricator MCP For?
This is for software engineers and project leads who spend too much time context-switching. If you're constantly jumping between your IDE, Jira, and GitHub just to track a single bug fix or review history, this server saves you clicks.
You use it to check task details or update code revisions directly from your IDE. You don't have to switch windows just to move a ticket status.
You track commit history and repository status across the entire organization without running manual queries in multiple dashboards.
You audit task histories using natural language queries to understand why a sprint is stalled or who was responsible for a critical step months ago.
What Changes When You Connect
- Audit a task's full lifecycle. Instead of digging through multiple logs to see who changed what and when, use
maniphest_gettasktransactionsto get the entire history in one query. - Keep your focus where it belongs: coding. You can update ticket status or assign ownership using
maniphest_editwithout ever leaving your IDE environment. - Find code changes fast. Need to see if a specific commit exists? Run
diffusion_commit_searchto check the exact ID against all repositories instantly. - Streamline reviews. Don't guess who worked on what. Use
differential_revision_searchto query and find related code revisions immediately. - Know your user context. The
user_whoamitool confirms exactly which account you are operating under, preventing accidental commits or status changes. - Targeted searching saves time. Instead of manually filtering a massive task list, use complex filters with
maniphest_searchto find only 'open' tasks assigned to specific teams.
Real-World Use Cases
The Stalled Feature Flag
A Project Lead notices a feature ticket (T901) hasn't moved in two weeks. They ask their agent: 'Show me the full transaction history for T901.' The agent runs maniphest_gettasktransactions and reports that the last activity was a comment on Friday, indicating no one has picked it up since.
The Code Review Bottleneck
An Engineer needs to check if a dependency commit exists. They ask: 'Search for commits related to library X.' The agent runs diffusion_commit_search and provides the exact SHA hashes, allowing them to continue their work without needing access to the full Git UI.
Updating Task Status Mid-Flow
A developer finishes a small bug fix. Instead of opening Phabricator and changing the status manually, they tell their agent: 'Mark T452 as ready for review.' The agent calls maniphest_edit and updates the task status immediately.
Onboarding Context Check
A new team member needs to know who owns a certain module. They ask: 'Who is responsible for the Auth service?' The agent runs user_search to find key contributors, giving the team lead immediate names and IDs.
The Tradeoffs
Asking too broad of a question
Typing 'Tell me about all tasks' or 'What is the repo status?' This yields an unusable wall of text because the server can't guess your constraints.
→
Always constrain your request. Instead, ask: 'Search for open Maniphest tasks assigned to group Alpha with priority high.' Use maniphest_search and provide specific filters.
Assuming a single API endpoint does everything
Trying to use the same command to find a task, edit it, and check its history. These are separate operations.
→
Break down the request. First, use maniphest_search to locate the task ID. Second, use that ID with maniphest_gettasktransactions for history. Third, use maniphest_edit if you need to change something.
Forgetting credentials
The agent reports an 'Authentication Failed' error and stops. You can't fix the problem by just rewriting the prompt.
→
Verify your connection first. Use user_whoami to confirm that your API token is active and linked to the correct Phabricator account before running any write operations.
When It Fits, When It Doesn't
Use this server if your workflow requires coordinating three different dev functions: task tracking (Maniphest), code review status (Differential), AND repository data (Diffusion). This is ideal for complex, multi-stage development processes.
Don't use it if you only need to look up simple user details or perform isolated actions. For example, if you just want a list of all users, user_search works. But if you need that list and then need to check their last task status and then find commits related to them, this server handles the entire chain.
It's critical to know its boundaries: it manages data within Phabricator; it doesn't replace GitHub or Jira entirely—it connects to what's already there. If your process relies on a system not connected via Conduit API, you need another tool type instead.
Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by Phabricator. All third-party trademarks, logos, and brand names are the property of their respective owners. Their use on this website is strictly for informational purposes to identify service compatibility and interoperability.
VINKIUS INFRASTRUCTURE
Cloud Hosted
Managed infra
V8 Isolated
Sandboxed per request
Zero-Trust Proxy
No stored credentials
DLP Enforced
Policy on every call
GDPR Compliant
EU data residency
Token Compression
~60% cost reduction
Works with Claude, ChatGPT, Cursor, and more
The Model Context Protocol standardizes how applications expose capabilities to LLMs. Instead of operating in isolation, your AI gains direct access to external platforms, live data, and real-world actions through secure, standardized connections.
This server provides 10 capabilities that interface natively with Claude, ChatGPT, Cursor, and any MCP client. No middleware. No custom integration required.
Available Capabilities
Jumping between tabs to track down a single bug fix is exhausting.
Right now, tracking progress means bouncing between the task board (to see status), the code review page (to check diffs), and maybe a separate Git UI (to find commits). You copy ticket IDs here, paste them there. It's slow, and you lose context.
With this MCP server, your AI agent does it all in one conversation. Ask for task T123's history, and the agent runs `maniphest_gettasktransactions`. The result is clean, immediate, and actionable—no switching tabs required.
Using `differential_revision_search` makes code reviews conversational.
Before this tool, finding a specific diff meant knowing the exact revision ID and navigating deep into Differential's UI. It was always manual data entry, prone to error.
Now you just ask your agent: 'Find revisions for commit X related to Y.' The agent uses `differential_revision_search` and pulls the necessary details right back to you. Simple.
Common Questions About Phabricator MCP
How do I find an open task assigned to me using maniphest_search? +
You use maniphest_search with specific constraints. You need to pass filters like 'status=open' and 'assigned_to={user}' so the search engine knows exactly what you mean.
Can I update a task status using maniphest_edit? +
Yes, that's one of its main jobs. You call maniphest_edit and tell it which task ID to change and the new status or owner you want.
What is the difference between diffusion_repository_search and diffusion_commit_search? +
Repository search finds containers—the overall project space. Commit search digs deeper, finding specific snapshots (SHA hashes) within those projects.
Which tool should I use to see who last worked on a task? Is it user_search or maniphest_gettasktransactions? +
Use maniphest_gettasktransactions. This tool gives the full timeline, showing which specific user made the last status change or comment.
What credentials do I need to use `diffusion_repository_search`? +
You must provide a valid Phabricator Instance URL and a Conduit API Token. The token needs read permissions for the repository section; otherwise, the search will fail.
How do I check my current credentials using `user_whoami`? +
The tool immediately returns your authenticated details, including your user ID and email. This confirms if your agent has permission to perform actions on your behalf.
What format should I use for complex filters when calling `maniphest_search`? +
You must pass constraints using a structured dictionary or JSON payload. The API requires specific syntax keys for project IDs, status codes, and date ranges to function correctly.
If I use `differential_revision_edit`, how does the system handle conflicting revisions? +
The server checks for existing conflicts and returns a dedicated error code. You'll need to manually resolve the conflict—often by merging changes—before re-attempting the edit via your AI client.
Can I update the status of a Maniphest task using this server? +
Yes. You can use the maniphest_edit tool. You'll need to provide the task identifier (like T123) and a transaction object specifying the 'status' type and the desired value.
How do I see the full history of changes for a specific task? +
Use the maniphest_gettasktransactions tool with the task ID (e.g., 'T123'). It will return a chronological list of all edits, comments, and status changes made to that task.
Can I search for code revisions that are currently open? +
Absolutely. Use differential_revision_search and pass a constraint for statuses like 'open' or 'needs-review' to filter the results to active code reviews.
Use it with your favorite AI tools
Connect this server to Cursor, Claude, VS Code, and more.
More in this category
ContentStack (Management)
Grant your AI agent read-write capabilities using ContentStack CMA. Update, publish, and manage contents, schemas, and environments from chat.
Tenderly (Ethereum Dev Platform)
Simulate Ethereum transactions, create Virtual TestNets, and monitor on-chain events directly from your AI agent.
Logto (Auth Platform)
Manage users, roles, and organizations in your Logto auth tenant directly from your AI agent.
You might also like
Commodities
Access real-time futures prices for 30+ commodities — gold, silver, oil, gas, wheat, coffee, copper, and more via Commodities-API.
SEC EDGAR Full — The Ultimate Free Bloomberg Alternative for AI Agents
The definitive SEC EDGAR Mega-Server: 13 tools spanning company lookup (8,000+ tickers), all filing types (10-K, 10-Q, 8-K), insider trading (Form 4), XBRL financial data extraction (revenue, income, assets, EPS), industry-wide comparison frames, and full-text search across every document ...
OpenWeather Agro
Access agricultural weather and satellite data via OpenWeather — monitor NDVI, soil temperature, evapotranspiration, frost risk, and GDD from any AI agent.