chatwatch
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| CHATWATCH_CDP | No | use a Chrome you already run with --remote-debugging-port | |
| CHATWATCH_HOME | No | config, state, events, drafts, API key, Chrome profile | ~/.chatwatch |
| CHATWATCH_POLL | No | seconds between checks | 4 |
| CHATWATCH_PORT | No | page and local API port | 8766 |
| CHATWATCH_CHROME | No | path to the Chrome executable | |
| CHATWATCH_CDP_PORT | No | DevTools port of chatwatch's Chrome | 9334 |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| teams_list_chatsA | List the Teams chats with their title, id, last message preview, unread flag and settings (watch, send mode off|draft|auto, handler). refresh=true reads the page again. |
| teams_read_messagesA | Read the last n messages of a chat (title, title fragment or id) without marking it read. |
| teams_wait_for_messagesA | New incoming messages in watched chats after event number after_seq, waiting up to wait_seconds (max 60) for one. Without after_seq, only messages from now on. Returns events and seq: pass seq as after_seq next time. With chat, other chats' messages are skipped and the answer may be empty before wait_seconds. |
| teams_send_messageA | Send text, and optionally local files (absolute paths), to a chat. The chat's send mode decides: auto sends now (status sent), draft saves it for a person to approve (status draft), off refuses. not_sent: nothing went out, you may try again. maybe_sent: do not resend; ask a person to check the chat. Sending opens the chat, which marks it read. |
| teams_get_attachmentsA | Download the pictures and files of the given messages (mid values) of a chat into a new local folder; returns the paths. |
| teams_list_draftsC | Replies waiting for a person to send or discard them on the chatwatch page. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 6 tools
Most tools target clearly different operations: listing chats, reading history, waiting for new events, sending, fetching attachments, and listing drafts. The closest pair is teams_read_messages vs teams_wait_for_messages, but the descriptions make the distinction (historical snapshot vs polling with after_seq/seq cursors) explicit enough that an agent can choose correctly.
Every tool uses the same teams_ prefix followed by a verb_noun snake_case pattern (list_chats, read_messages, send_message, get_attachments, list_drafts). No mixing of camelCase or inconsistent verb styles.
Six tools is well-scoped for a Teams chat bridge: one per lifecycle step (discover chats, read, poll, send, fetch attachments, review drafts) with no redundant entries. Nothing feels padded or missing at the count level.
Core read/poll/send/attachment/draft-listing flows are covered, but teams_list_chats exposes per-chat settings (watch, send mode, handler) with no tool to change them, and drafts can only be listed, never approved or discarded programmatically. Sending also opens/marks a chat read with no standalone mark-read/unread operation, forcing workarounds or human action on the chatwatch page.