zaungast
zaungast is a read-only, offline MCP server that queries your local Microsoft Teams disk cache without any cloud API, credentials, or network calls.
list_conversations– Browse your Teams sidebar: newest conversations, filtered by kind (1:1, group, channel, meeting), participant, title, or time range.read_messages– Read a conversation's messages in chronological order; supports cursor-based pagination, time-windowed reading, centering around a specific message, and reaction details.search– Full-text search across all messages with filters for sender, conversation, kind, date range, mentions, attachment presence, and exclusions.list_events– Retrieve metadata for calendar meetings and appointments (no join URLs).list_calls– Browse call history: 1:1 and group calls with direction, duration, missed status, and recording pointers.top_topics– Discover trending or distinctive topics over a time window (1d/7d/30d), optionally scoped to a person or conversation, with an example message per topic.find_person– Resolve a name or nickname to a canonical person with their stable handle, message count, and last-contact time; supports browsing the full roster.describe_schema– Inspect the raw Teams IndexedDB structure and propose field mappings — useful when a Teams update changes the database layout.
It is strictly read-only, requires no credentials or tokens, does not use the Microsoft Graph API, and produces token-economical output designed for efficient use with AI/coding agents.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@zaungast find messages from Anna about the deploy"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
zaungast
zaungast (German: someone who watches over the fence without joining in) is a read-only, offline MCP server for Teams — search chats, read conversations, surface trending topics, and find people straight from the local on-disk cache, with no Graph API, no cloud, and no credentials, and token-economical output for coding agents (Claude Code, Claude Desktop, …).
The new Teams client stores your chats in a local on-disk database; zaungast reads a copy of it directly and serves it over MCP, so your agent can pull in Teams context — "what was decided about the release date", "catch me up on a channel I muted", "what's my team been discussing this week" — without you copy-pasting, and without any cloud API.
Local & offline — reads the on-disk Teams cache. No MS Graph API, no network calls.
No credentials — nothing to log in to, no tokens, no permissions to grant.
Read-only & safe — the Teams files are only ever read/copied, never written, locked, or modified. It cannot corrupt your Teams data.
Token-economical — every tool returns compact, shaped output, never bulk dumps.
Zero-config — auto-discovers the local Teams database; just register and go.
⚠️ Windows & macOS (new Teams / WebView2), Node.js ≥ 22.5. Both use the same Chromium/WebView2 store, so the reader is identical; the local database is auto-discovered on each, or point
TEAMS_LEVELDB_DIRat it manually — see requirements. Not affiliated with or endorsed by Microsoft. It reads your own local data on your own machine.
Installation
npx -y zaungast fetches and runs the server, so registering it in your MCP client is the
whole install. No environment variables are needed in the common case — the local Teams
database is auto-discovered.
Claude Code
Normal case — the Teams database is auto-discovered, so no variables are needed:
claude mcp add zaungast -- npx -y zaungastAdd --scope user to make it available in every project. Verify with claude mcp list
(expect zaungast … ✓ Connected), then open a new session.
Only if auto-discovery fails or you have multiple profiles, set TEAMS_LEVELDB_DIR
explicitly (the -e flag goes before the --) — see
finding the Teams database folder:
claude mcp add zaungast \
-e TEAMS_LEVELDB_DIR="C:\Users\me\AppData\Local\Packages\MSTeams_8wekyb3d8bbwe\LocalCache\Microsoft\MSTeams\EBWebView\WV2Profile_tfw\IndexedDB\https_teams.microsoft.com_0.indexeddb.leveldb" \
-- npx -y zaungastCodex
Add to ~/.codex/config.toml (or run codex mcp add zaungast -- npx -y zaungast):
[mcp_servers.zaungast]
command = "npx"
args = ["-y", "zaungast"]
# The Teams database is auto-discovered, so no path is needed. To set or override it,
# delete the "#" below and point it at your …indexeddb.leveldb folder:
# env = { TEAMS_LEVELDB_DIR = "/full/path/to/https_teams.microsoft.com_0.indexeddb.leveldb" }Claude Desktop / other clients
Add to your client's MCP config (claude_desktop_config.json, .mcp.json, …):
{
"mcpServers": {
"zaungast": { "command": "npx", "args": ["-y", "zaungast"] }
}
}If you want to set or override a variable, use this form instead:
{
"mcpServers": {
"zaungast": {
"command": "npx",
"args": ["-y", "zaungast"],
"env": {
"TEAMS_LEVELDB_DIR": "/full/path/to/https_teams.microsoft.com_0.indexeddb.leveldb"
}
}
}
}TEAMS_LEVELDB_DIR is optional — the database is auto-discovered. Set it only if discovery
fails or to pin a specific profile. From-source setup and other clients are covered in the
installation docs.
Related MCP server: Teams Messenger MCP App
Tools
Tool | What it does |
| Your Teams sidebar — newest conversations, or filter by kind/participant/title/time. |
| Browse one conversation's messages in story order (window / date-range / page back). |
| Read one channel reply-chain in full (root + all replies). |
| One message in full — complete untruncated body + reactions, by conversation + |
| Full-text search + filters (from, in, kind, |
| Calendar meetings & appointments (forward window by default); metadata-only, join-URLs never exposed. |
| Call history — 1:1/group calls with direction, duration, missed, and recording pointers. |
| Rank distinctive/trending topics over a window, vs your baseline, with an example each. |
| Resolve a name/nickname to a canonical person + handle, with contact stats. |
| Recovery tool: propose a field mapping when a Teams update changes the DB layout. |
Full reference: tools documentation.
Environment variables
All optional — zaungast works with no configuration. Pass them via your MCP client's env
block (as above).
Var | Default | Notes |
| auto-discovered | Absolute path to the |
|
| Refresh mode: |
| unset | Read a static copy of the database directly (offline analysis); skips discovery and live refresh. |
|
| Ingest engine: |
TEAMS_LEVELDB_DIR points at the folder ending in .indexeddb.leveldb (the one holding
CURRENT, MANIFEST-*, *.ldb/*.log) — not its parent IndexedDB directory:
# Windows
C:\Users\<you>\AppData\Local\Packages\MSTeams_8wekyb3d8bbwe\LocalCache\Microsoft\MSTeams\EBWebView\<profile>\IndexedDB\https_teams.microsoft.com_0.indexeddb.leveldb
# macOS
/Users/<you>/Library/Containers/com.microsoft.teams2/Data/Library/Application Support/Microsoft/MSTeams/EBWebView/<profile>/IndexedDB/https_teams.microsoft.com_0.indexeddb.leveldb<profile> is usually WV2Profile_tfw. Use the full absolute path in JSON configs (e.g. Claude
Desktop) — environment placeholders like ~ or %LOCALAPPDATA% aren't expanded there.
Privacy & safety
Stays local — no network calls; reads data already on your machine and serves it to your local agent over stdio.
Cannot harm Teams — no code path writes to, locks, or memory-maps the Teams directory; only read-and-copy.
Images/files are URL-only — chat images live in Teams' cloud behind auth; zaungast notes that an attachment exists but never fetches it and never handles credentials.
More in the privacy & safety docs.
Documentation
Full documentation — tools reference, how it works, configuration, privacy, troubleshooting, and development — lives at https://zaungast.readthedocs.io/.
License
Apache 2.0 © Mikael Beyene. Not affiliated with or endorsed by Microsoft. "Microsoft Teams" is a trademark of Microsoft Corporation; this project only reads your own local data.
Available Tools
6 toolsdescribe_schemaDescribe / recover schemaA
Inspect the raw Teams IndexedDB stores and PROPOSE a field mapping. Use when tools report the schema is unrecognized (after a Teams update), or to inspect the DB structure. Proposes only — applies nothing; a human verifies the proposal and saves it as a new schema version.
Inputs: limit.
find_personFind a personA
Resolve a name/nickname fragment to a person's canonical name and stable p:handle, with message count and last-contact time. Use when a name is ambiguous or unrecognized, or when you need contact stats or a p:handle. For ordinary filtering, just pass a name substring directly to search/read_messages 'from'/'participant' — don't call this first. Omit query to scan the roster (most-talked-to first).
Inputs: n, query.
list_conversationsList conversationsA
Your Teams sidebar: the newest N conversations (default), or filter by kind/participant/title/time. Note: this reads the LOCAL Teams cache — history is the synced slice on this device, not the full server archive.
Inputs: n, kind, query, since, +2 more.
read_messagesRead a conversationA
Read one conversation's messages in STORY ORDER (oldest→newest). Target by handle (c:xxxx) or title/participant substring. Page back with the returned older: cursor, or center on a message with around:. Note: this reads the LOCAL Teams cache — history is the synced slice on this device, not the full server archive.
Inputs: conversation (required), limit, since, until, +3 more.
searchSearch messagesA
Full-text search across all messages with filters. Empty query = filtered browse. from/in accept display-name / title substrings or handles. mentions_me finds messages that @mention you. Note: this reads the LOCAL Teams cache — history is the synced slice on this device, not the full server archive.
Inputs: in, from, kind, limit, +6 more.
top_topicsTrending topicsA
Distinctive/trending topics over a window (vs your baseline), overall or scoped to a person/conversation. Returns each topic with an exemplar message. Note: this reads the LOCAL Teams cache — history is the synced slice on this device, not the full server archive.
Inputs: n, scope, since, until, +3 more.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 6 tools
Each tool has a clear and distinct purpose: schema inspection, person resolution, conversation listing, message reading, full-text search, and topic extraction. No overlap in functionality.
All tool names use consistent snake_case and follow a verb_noun pattern (describe_schema, find_person, list_conversations, read_messages, top_topics), with 'search' as a minor deviation but still common and predictable.
6 tools is well-scoped for a Teams data access server, covering essential read and analysis operations without being overwhelming or sparse.
The tool set covers reading and analyzing local cache data comprehensively (schema, people, conversations, messages, search, topics). Minor gaps include lack of write operations and detailed conversation metadata retrieval, but this aligns with the stated read-only purpose.
Maintenance
Related MCP Connectors
An MCP server that provides read access to your cloud storage providers, bank accounts and more.
Search your AI chat history (ChatGPT, Claude, Codex) from any MCP client. Remote, private, read-only
Your own WhatsApp as an MCP server: read, search and send from any MCP client.
Read-only MCP server for ClassQuill, a tutoring-business-management platform.
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceA Model Context Protocol server that bridges Microsoft Teams and MCP-compatible clients, enabling chat integration, message handling, and advanced search capabilities without REST API endpoints.1-
- FlicenseNot gradedqualityDmaintenanceA Model Context Protocol server that bridges Microsoft Teams with MCP-compatible clients (LLMs, agentic frameworks, CLI), enabling chat integration, message search, and event streaming without traditional REST API endpoints.-
- AlicenseAqualityBmaintenanceLocal-first, read-only MCP server for searching and retrieving cited evidence from archived Telegram chats, including transcripts and media metadata.53MIT
- AlicenseBqualityCmaintenanceReads the local Microsoft Teams (v2) message cache and exposes it over MCP, enabling AI assistants to list accounts/conversations and search/read messages without Graph API or network.88MIT