Skip to content

USAGE.md

Latest commit

 

History

History
207 lines (133 loc) · 5.26 KB

USAGE.md

File metadata and controls

207 lines (133 loc) · 5.26 KB

Usage Guide

This plugin is designed for AI-native interaction — you talk to Claude Code in natural language, and slash commands handle the API calls behind the scenes. You don't need to memorize flags or syntax.

Getting Started

> /slack init

Claude will walk you through extracting tokens from your browser and writing the .env file. Then verify:

> /slack status

Everyday Examples

Search for messages

> /slack search database connection timeout
> /slack search PROJ-12345
> /slack search deploy failure --channel team-platform
> /slack search "memory leak" --sort timestamp

Or just ask Claude directly:

> find Slack messages about connection pool exhaustion
> has anyone discussed the v2.5 upgrade issue in Slack?
> search for messages mentioning ticket PROJ-456

Read a channel

> /slack read team-backend
> /slack read team-backend --since 3
> /slack read C0123ABCDEF --limit 10

Or conversationally:

> what's been happening in #team-backend?
> show me the last 3 days from #incidents

Follow a thread

When you see an interesting message with replies, read the full thread:

> /slack read team-backend --thread 1773943155.346989

Thread timestamps (ts) are shown in search results and channel history. Claude will often suggest reading a thread when it spots one with replies.

Send a message

> /slack send #team-backend "Deploy is complete, all tests passing"
> /slack send team-backend "Looks good to me" --thread 1234567890.123456

Or conversationally:

> send a message to #team-backend saying the deploy is done
> reply to that thread saying I'll look into it

Note: Claude will always show you the message preview and ask for confirmation before sending.

List channels

> /slack channels
> /slack channels platform
> /slack channels infra

On enterprise workspaces, the channel list builds gradually from search results. The more you search, the more channels are discovered and cached.

Working with Enterprise Slack

On enterprise Slack instances, the channel listing API is often restricted. This tool handles it automatically:

  1. Search first — /slack search always works and discovers channel IDs
  2. Channel names resolve via search — when you use a channel name, the tool searches in:#channel-name to find the ID
  3. IDs are cached — once a channel is discovered, it's remembered for future use
  4. Direct IDs always work — use channel IDs from Slack URLs (e.g., C0123ABCDEF)

Typical workflow:

> /slack search deploy failure
  (results show channels with IDs: #team-platform (C012ABC), #incidents (C034DEF))

> /slack read C012ABC --limit 10
  (read directly by ID)

> /slack read team-platform
  (now works by name — the ID was cached from search)

Combining with Other Tools

Slack + Jira

Search Slack for discussions about a Jira ticket:

> /slack search PROJ-789

  (Claude finds Slack threads discussing this ticket)

> /jira start PROJ-789

  (pick up the ticket and start working)

Or discover Jira tickets mentioned in Slack conversations:

> /slack read team-backend --thread 1773943155.346989

  Claude: "This thread mentions PROJ-789 and INFRA-234.
           Want me to look them up?"

> yes, check their status

Slack + Support Cases

Find Slack context for a support case:

> /rh-case show 12345678

  (read the case details)

> /slack search 12345678

  (find internal Slack discussions about this case)

> /slack search "connection timeout" database pool

  (search for the error message from the case)

Three-Way Triage

A complete investigation using all three tools:

> /slack search "connection pool exhaustion"

  (find the original discussion)

> /slack read team-backend --thread 1773943155.346989

  (read the full thread — discover PROJ-789 and case 12345678)

> /rh-case show 12345678 --comments

  (read the support case)

> /rh-case search --type kcs connection pool timeout

  (check if there's a knowledge base article)

  Claude: "Based on the Slack thread, Jira ticket, and support case:
           The issue is caused by a connection leak in v2.5.1.
           Fix is in PR #841, release in progress.
           Workaround: increase pool max_connections to 50."

Workflow Example: Morning Triage

> /slack search "from:me" --sort timestamp --count 5

  (check your recent messages)

> /slack read team-standup --since 1

  (what did the team post yesterday?)

> /jira status

  (your current Jira task)

> /rh-case list urgent

  (any hot support cases?)

Tips

  • Be conversational — You don't need exact command syntax. Claude understands intent like "search Slack for deploy issues" or "read the backend channel."
  • Follow up — After any search or read, ask Claude to summarize, compare messages, draft a response, or dig into a thread.
  • Channel IDs are your friend — On enterprise Slack, use channel IDs from search results or Slack URLs. They're more reliable than names.
  • Cross-reference — When Claude spots Jira tickets or case numbers in Slack messages, ask it to look them up. That's the power of having all three tools in one session.
  • Stay in context — Once messages are loaded, Claude remembers them. You can ask follow-up questions without re-searching.