Skip to main content
Glama

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault
SLACK_TOKENNoSlack xoxc token from your browser session. Required on Windows/Linux if not using auto-extraction (macOS).
SLACK_COOKIENoSlack xoxd cookie from your browser session. Required on Windows/Linux if not using auto-extraction (macOS).

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

CapabilityDetails
tools
{}
prompts
{}
resources
{}

Tools

Functions exposed to the LLM to take actions

NameDescription
slack_token_statusA

Check token health, age, auto-refresh status, and cache stats

slack_health_checkA

Check if Slack tokens are valid and show authentication status

slack_refresh_tokensA

Force refresh tokens by extracting from Chrome (macOS; requires having signed into Slack in Chrome)

slack_list_conversationsA

List all DMs and channels with user names resolved. Uses cached DMs by default for speed.

slack_conversations_historyC

Get messages from a channel or DM with user names resolved

slack_get_full_conversationB

Export FULL conversation history with all messages, threads, and user names. Can save to file.

slack_search_messagesC

Search messages across the Slack workspace

slack_users_infoB

Get detailed information about a Slack user

slack_send_messageC

Send a message to a channel or DM

slack_get_threadC

Get all replies in a message thread

slack_list_usersB

List all users in the workspace

slack_add_reactionA

Add an emoji reaction to a message

slack_remove_reactionB

Remove an emoji reaction from a message

slack_conversations_markA

Mark a conversation as read up to a specific message timestamp

slack_conversations_unreadsA

Get channels and DMs with unread messages, sorted by unread count (highest first)

slack_users_searchA

Search workspace users by name, display name, or email. Case-insensitive partial match.

slack_workflow_saveA

Save or update a workflow profile that binds a workflow_kind (support_inbox | incident_room | exec_brief | product_launch_watch | custom) to channels, priority people, retention mode, and summary cadence. Stored locally at ~/.slack-mcp-workflows.json. slack_catch_me_up reads the profile by name and returns evidence shaped by its workflow_kind.

slack_workflowsA

List all saved workflow profiles from ~/.slack-mcp-workflows.json. Optionally filter by workflow_kind. Returns profile_name, channels, priority_people, retention_mode, summary_cadence, structured_keys, created_at, updated_at.

slack_catch_me_upA

Catch up on a saved workflow profile. Reads the profile's channels (or everything currently unread if the profile names none), pulls messages since the cadence window or an explicit since, expands active threads, and returns structured evidence: which threads are unanswered and for how long, what the profile's priority people said or were pinned on, and which conversations moved most. Runs locally against your own session — no hosted account, no server-side model. It also returns a continuity block — commitments someone made, asks nobody has claimed, and questions still unanswered — built from phrase and reply-count rules over the same messages, with the matched text and a permalink on every row so each line can be opened and checked. The response carries an output_contract naming the keys to compose for this workflow_kind; write the summary from the returned signals, continuity and conversations, citing conversation names and timestamps.

slack_session_reportA

What this server process has actually done, as counts: how many messages it read, how many of those were written by authors outside this workspace broken down by origin, how many sends were attempted, and how many were held. A provenance label is a claim about one message; this is the claim about the whole session, and it is the only way to check after the fact whether outside-authored text reached the model and whether anything tried to send on the back of it. Counts only — no message text, no channel name, no user id is recorded. The receipt and the strict hold belong to this server process, so over HTTP every client of the process shares them.

Prompts

Interactive templates invoked by user choice

NameDescription
search-recentSearch workspace for messages from the past week
summarize-channelGet recent activity from a channel for summarization
find-messages-fromFind all messages from a specific user

Resources

Contextual data attached and managed by the client

NameDescription
Workspace InfoCurrent workspace name, team, and authenticated user
ConversationsList of available channels and DMs

TDQS

A3.5/5.0

Scored across 20 tools

Disambiguation4/5

Most tools target clearly distinct Slack resources and actions, but slack_health_check, slack_token_status, and slack_refresh_tokens form an overlapping token-management cluster, and slack_conversations_history vs slack_get_full_conversation can be confused. Descriptions largely disambiguate, so an agent can usually choose correctly.

Naming Consistency4/5

All tools use the slack_ prefix and snake_case, which is highly predictable, but internal ordering alternates between verb_noun (list_conversations, send_message) and noun_verb (users_info, conversations_history). This mirrors Slack's API inconsistently but remains readable.

Tool Count3/5

20 tools is in the heavy range for an MCP server, and some peripheral clusters (token management, workflow profiles, session reporting) could potentially be consolidated. However, Slack is a broad domain, so most tools earn their place.

Completeness4/5

Core Slack workflows are covered: reading, sending, searching, threads, reactions, user lookup, and conversation management. Gaps exist for editing/deleting messages, file uploads, and channel administration, but agents can work around these for the server's monitoring and catch-up focus.

Maintenance

ActivityActive
ResponsivenessResponsive