Slack Mastery Guide
Architecting the Primary Receptor
This guide is authored for the discerning engineer who demands the highest level of integration. The HotPlex Slack Adapter is not just a bot; it is a high-performance cognitive bridge that leverages Slack's Block Kit to manifest AI agency with unprecedented clarity.
⚡ Rapid Manifest Deployment
The most refined path to integration is via the App Manifest. This allows you to orchestrate dozens of permissions and features in a single, atomic declaration.
TIP
For visual step-by-step instructions on creating your Slack App and obtaining tokens, refer to the Slack Beginner Guide. For a deep reference on all configuration parameters, see the Core Configuration Manual.
display_information:
name: HotPlex
description: High-Performance AI Agent Receptor
background_color: "#000000"
features:
bot_user:
display_name: HotPlex
always_online: true
slash_commands:
- command: /reset
description: Re-initialize the agent context (Cold Start)
- command: /dc
description: Terminate background long-running processes
oauth_config:
scopes:
bot:
- app_mentions:read
- chat:write
- chat:write.public
- reactions:write
- im:history
- channels:history
- groups:history
- files:write
- commands
settings:
event_subscriptions:
bot_events:
- app_mention
- app_home_opened
- message.channels
- message.groups
- message.im
interactivity:
is_enabled: true
socket_mode_enabled: true✨ The Visual Language of Agency
1. Native Assistant Status (2026 AI-Native UX)
HotPlex leverages Slack's Assistant Threads API for real-time progress indication—a premium experience that keeps chat history clean:
| Status | Visual | Description |
|---|---|---|
in_progress | 🧠 Thinking... | Agent is reasoning or planning |
tool_use | 🔧 Using tool... | Agent is executing a tool/command |
tool_error | ⚠️ Tool error | Tool execution failed |
This replaces legacy "typing indicator" and bubble simulations with native Slack functionality.
2. WAF Closed Loop (Danger Block)
When the engine's Regex WAF detects a dangerous command, execution pauses for human approval:
User: "Delete all files with rm -rf /"
→ WAF blocks → [Interactive Danger Card appears]
→ User clicks "Confirm" or "Cancel"
→ Execution resumes or terminatesThe interactive confirmation uses Slack Block Kit with danger style buttons for clear visual hierarchy.
3. Space Folding (Threaded Output)
High-volume tool outputs are automatically folded into thread replies:
- Main channel shows only tool call summary
- Full output lives in thread (expandable)
- Prevents channel pollution while preserving audit trail
4. The Pulse of Progress (Reactions)
The agent communicates its internal state via a subtle language of emojis:
- 📥 (
:inbox_tray:): The request has been queued and acknowledged in the session pool. - 🧠 (
:brain:): The engine is performing deep cognitive reasoning or planning. - 🔨 (
:hammer_and_wrench:): The agent is actively executing tools, shell commands, or background processes. - ⏳ (
:hourglass_flowing_sand:): Execution is suspended, awaiting human-in-the-loop (HITL) authorization or input. - ✅ (
:white_check_mark:): The interaction loop has reached a successful resolution. - ❌ (
:x:): An error was encountered during the process. Check the Display area for details. - 🚫 (
:no_entry_sign:): A security block or safety policy violation was detected.
2. Structural Interaction (6 Zones Mapping)
Every agent message sequence is divided into atomic Zones to ensure clarity and avoid history bloat:
- Zone 0 (Status): Ephemeral indicator (
Initializing...) destroyed upon further action. - Zone 1 (Thinking): Real-time context limited to a scrolling window of the latest 64-characters with a 1-second cadence throttle to maintain UI sanity.
- Zone 2 (Action): High-visibility
tool_use/tool_resultcards strictly capped to a sliding window of 2 messages. Older actions are automatically purged. - Zone 3 (Display): The final intellectual artifact, auto-paginated using divider blocks every ~3500 bytes for unlimited fluid streaming.
- Zone 4 (Interaction): Blocking wait state containing interactive prompt elements (HITL approvals, questions).
- Zone 5 (Summary): Session-end statistics indicating token and duration metrics.
5. Absolute Black Hole Policy
HotPlex implements a strict noise-filtering philosophy:
- Silences system-level logs within the integration layer
- Drops redundant user reflections (duplicate confirmations)
- Filters boilerplate AI responses before they reach Slack UI
This ensures users see only meaningful content, not implementation artifacts.
🛠️ Deterministic Configuration
Fine-tune the behavior of your receptor in chatapps/configs/slack.yaml:
bot_user_id: Mandatory. Ensures the receptor recognizes its own identity.dm_policy: Choose betweenalloworpairing(restricts DMs to known users).group_policy: Control whether the agent listens to all chatter or only explicit @mentions.
🏠 App Home Capability Center (v0.25.0+)
HotPlex supports Slack App Home - an interactive form-based interface for predefined AI tasks. Users can trigger specialized workflows directly from the Slack home tab.
Features
| Feature | Description |
|---|---|
| Code Review | Submit PR URLs for AI-powered code review |
| Debugging | Paste error messages for analysis and fixes |
| Documentation | Generate docs from code or specs |
| Testing | Create test cases from implementation |
| Refactoring | Get AI suggestions for code improvements |
Configuration
Enable in your Slack config:
# configs/base/slack.yaml
assistant:
app_home:
enabled: true
capabilities_path: ./capabilities.yamlCapabilities YAML
Define available capabilities in capabilities.yaml:
capabilities:
- id: code_review
name: Code Review
description: Submit a PR URL for AI-powered code review
form:
- name: pr_url
type: text
label: Pull Request URL
required: true
- name: focus
type: select
label: Review Focus
options:
- Security
- Performance
- Best Practices
- General"Integrate not just for function, but for the experience of collaboration." — The HotPlex Team