Delv
CommunityAbandoned· 1.2y4.3by Jagan Shanmugam

Mattermost MCP Host

Connects to Mattermost teams, channels, and messages, running as a bot integrated into threads.

C
Safety & Trust

Delv Safety Grade: C

Score 58/100 · assessed 2026-04-28

Maintainer40
Permissions65
Supply chain45
Transparency70
Incidents100

This community MCP server connects Claude to self-hosted Mattermost instances as a bot, enabling message reading and posting across teams and channels. The maintainer appears to be a solo developer with limited visible track record. Installation requires cloning and running uv sync rather than a standard package manager, which increases supply chain friction. The server needs a Mattermost bot token with potentially broad permissions depending on how it's configured in Mattermost itself. Permissions are moderately scoped: it reads and writes messages within Mattermost but doesn't touch local filesystem or execute shell commands. The code is open source with reasonable documentation, though the project is relatively new with sparse community validation. No known security incidents. Suitable for teams already running Mattermost who understand the bot token's scope and trust the implementation, but the solo maintainer and non-standard install warrant caution for production deployments.

Lethal Trifecta (prompt-injection exposure)

TRIFECTA RISK
All three axes present. This server can read private data, ingest attacker-controlled content, and send data outbound. A poisoned input (a GitHub issue, an email, a webpage) can exfiltrate secrets via this chain. Only install with auditing; avoid on shared or cloud agents.
Private dataYes
Reads secrets, credentials, private files
Untrusted inputYes
Ingests web pages, PRs, issues, emails
External commsYes
Can send data outbound

Same shape as Slack. Self-hosted does not change the trifecta surface, only who the attacker has to be.

Green flags

  • Open source with clear repository and documentation
  • Targets self-hosted Mattermost, avoiding third-party SaaS dependencies
  • Scoped to messaging domain, no filesystem or shell access
  • No known security incidents or malicious behaviour

Red flags

  • Solo maintainer with limited public track record
  • Requires clone-and-build install rather than standard package registry
  • Bot token scope depends on Mattermost configuration, potentially very broad
  • New project with minimal community review or validation

Permissions requested

Outbound networkAccess secretsRead messagesSend messages
Assessed by Delv Editorial using public metadata. Grades are advisory and update as the ecosystem changes. They do not replace your own review of permissions and code before granting an agent access to sensitive systems.

Install

uv sync
Env vars needed: MATTERMOST_URLMATTERMOST_TOKEN

Review

This MCP server turns Claude into a Mattermost bot that can read channels, post messages, and participate in threads. You point it at your self-hosted Mattermost instance with a bot token, and suddenly your AI assistant can summarise long threads, answer questions in team channels, or monitor specific conversations. The implementation is straightforward: it exposes tools for listing teams, channels, posts, and creating new messages or replies. I'd reach for this if I'm already running Mattermost internally and want an AI presence that doesn't require SaaS integrations or third-party middleware. The self-hosted angle matters here. You control the data, the bot runs locally via MCP, and you're not routing internal chat through external APIs. For DevOps teams using Mattermost as their ops hub, this could slot into incident response workflows: Claude reads the incident channel, pulls context from previous messages, and posts summaries or suggested next steps. The setup requires a bot account in Mattermost and a personal access token. Once configured, the server connects as that bot user, so permissions matter. If your bot can't see a private channel, neither can Claude. That's sensible, but it means you need to think about bot membership upfront. The repo doesn't bundle pre-built binaries, so you're running from source with `uv sync`, which is fine if you're comfortable with Python tooling but adds friction for non-developers. Thread summarisation is the obvious win. Long async discussions in Mattermost can sprawl across days. Asking Claude to read the last fifty posts and extract action items actually works, and it's faster than scrolling. The bot can also post replies inline, so you can have it answer routine questions in support channels or acknowledge incident updates. The limitation is that it's read-and-write only. There's no webhook listener here, so Claude won't react to new messages automatically. You trigger it manually or script something external. This isn't for Slack shops, obviously, and if you're on Mattermost Cloud with strict API limits, you might hit rate issues. But for self-hosted Mattermost users who want a local AI assistant in their chat stack, this does exactly what it says with no fuss.
Verdict

Install this if you run Mattermost internally and want Claude to read, summarise, or post in channels without external dependencies. Skip it if you're on Slack, need real-time message reactions, or don't want to manage a bot token. It's a clean, focused tool for self-hosted teams.

Good at

  • Keeps all chat data on your self-hosted Mattermost instance, no external API routing required.
  • Thread summarisation and context retrieval work well for long async discussions.
  • Bot can post replies inline, useful for answering routine questions or acknowledging updates.
  • Straightforward permission model: bot sees what it's allowed to see, nothing hidden.
  • No SaaS dependencies or third-party middleware, just your Mattermost API and MCP.

Watch out

  • No webhook listener, so Claude won't react to new messages automatically, you trigger it manually.
  • Requires running from source with Python tooling, no pre-built binaries for quick setup.
  • Bot must be manually added to channels before Claude can access them, which adds admin overhead.
  • Not useful if you're on Slack or another platform, this is Mattermost-only.
  • API rate limits on Mattermost Cloud could cause issues with heavy usage.

Use cases

  • internal chat assistants
  • thread summarisation
  • self-hosted messaging
  • DevOps bots

Getting started

1. Create a bot account in your Mattermost instance and generate a personal access token with read/write permissions. 2. Clone the repo and run `uv sync` to install dependencies. 3. Set `MATTERMOST_URL` to your instance URL and `MATTERMOST_TOKEN` to the bot token, then add the server to your Claude Desktop config pointing to the repo path. 4. Restart Claude and ask it to list your Mattermost teams to verify the connection. 5. Watch out for bot channel membership: if the bot isn't in a channel, Claude can't see it, so add the bot to relevant channels first.

Works with

Claude DesktopClaude CodeCursor

Similar MCPs