Managing Stream Chat

Stream Chat Management Skill

Manage and analyze Stream Chat resources including channels, users, messages, and moderation.

API Conventions

Authentication

All API calls use API key and secret with server-side token, injected automatically.

Base URL

https://chat.stream-io-api.com

Core Helper Function

#!/bin/bash

stream_api() {
    local method="$1"
    local endpoint="$2"
    local data="${3:-}"

    if [ -n "$data" ]; then
        curl -s -X "$method" \
            -H "Authorization: $STREAM_AUTH_TOKEN" \
            -H "Content-Type: application/json" \
            -H "stream-auth-type: jwt" \
            "https://chat.stream-io-api.com${endpoint}?api_key=$STREAM_API_KEY" \
            -d "$data"
    else
        curl -s -X "$method" \
            -H "Authorization: $STREAM_AUTH_TOKEN" \
            -H "stream-auth-type: jwt" \
            "https://chat.stream-io-api.com${endpoint}?api_key=$STREAM_API_KEY"
    fi
}

Output Rules

• Target ≤50 lines per script output
• Use jq to extract only needed fields
• Never dump full API responses

Phase 1: Discovery

#!/bin/bash
echo "=== App Settings ==="
stream_api GET "/app" | jq '{name: .app.name, organization: .app.organization, push_notifications: .app.push_notifications}'

echo ""
echo "=== Channel Types ==="
stream_api GET "/channeltypes" \
    | jq -r '.channel_types | to_entries[] | "\(.key)\tmax_members:\(.value.max_message_length)\tautomod:\(.value.automod)"' \
    | head -15

echo ""
echo "=== Recent Channels ==="
stream_api POST "/channels" '{"filter_conditions": {}, "sort": [{"field": "last_message_at", "direction": -1}], "limit": 20}' \
    | jq -r '.channels[] | "\(.channel.id[0:20])\t\(.channel.type)\t\(.channel.member_count) members\t\(.channel.last_message_at[0:16] // "no messages")"' \
    | column -t | head -20

Phase 2: Analysis

Chat Health

#!/bin/bash
echo "=== Channel Summary ==="
stream_api POST "/channels" '{"filter_conditions": {}, "limit": 100}' \
    | jq '{
        total_channels: (.channels | length),
        total_members: (.channels | map(.channel.member_count // 0) | add),
        avg_members: (.channels | map(.channel.member_count // 0) | if length > 0 then add / length | floor else 0 end),
        by_type: (.channels | group_by(.channel.type) | map({(.[0].channel.type): length}) | add)
    }'

echo ""
echo "=== Active Channels (with recent messages) ==="
stream_api POST "/channels" '{"filter_conditions": {"last_message_at": {"$gt": "'$(date -d '24 hours ago' +%Y-%m-%dT%H:%M:%SZ)'"}}, "limit": 20}' \
    | jq -r '.channels[] | "\(.channel.id[0:20])\t\(.channel.type)\t\(.channel.member_count) members"' \
    | head -15

echo ""
echo "=== Inactive Channels (no messages in 30d) ==="
stream_api POST "/channels" '{"filter_conditions": {"last_message_at": {"$lt": "'$(date -d '30 days ago' +%Y-%m-%dT%H:%M:%SZ)'"}}, "limit": 20}' \
    | jq -r '.channels[] | "\(.channel.id[0:20])\t\(.channel.type)\t\(.channel.last_message_at[0:10] // "never")"' \
    | head -10

User & Moderation

#!/bin/bash
echo "=== User Query ==="
stream_api POST "/users" '{"filter_conditions": {}, "sort": [{"field": "last_active", "direction": -1}], "limit": 20}' \
    | jq -r '.users[] | "\(.id[0:20])\t\(.online)\t\(.last_active[0:16] // "never")\tbanned:\(.banned // false)"' \
    | column -t | head -15

echo ""
echo "=== Banned Users ==="
stream_api POST "/users" '{"filter_conditions": {"banned": true}, "limit": 20}' \
    | jq -r '.users[] | "\(.id[0:20])\t\(.ban_expires[0:16] // "permanent")"' | head -10

echo ""
echo "=== Flagged Messages ==="
stream_api GET "/moderation/flags/messages?limit=20" \
    | jq -r '.flags[] | "\(.message.id[0:16])\t\(.message.text[0:40])\t\(.user.id)\t\(.created_at[0:16])"' | head -10

Output Format

=== Stream Chat App: <name> ===

--- Channels ---
Total: <n>  Active (24h): <n>  Inactive (30d): <n>
By Type: messaging: <n>, livestream: <n>

--- Users ---
Total: <n>  Online: <n>  Banned: <n>

--- Moderation ---
Flagged Messages: <n>

Anti-Hallucination Rules

1. NEVER assume resource names — always discover via CLI/API in Phase 1 before referencing in Phase 2.
2. NEVER fabricate metric names or dimensions — verify against the service documentation or --help output.
3. NEVER mix CLI commands between service versions — confirm which version/API you are targeting.
4. ALWAYS use the discovery → verify → analyze chain — every resource referenced must have been discovered first.
5. ALWAYS handle empty results gracefully — an empty response is valid data, not an error to retry.

Counter-Rationalizations

Common Pitfalls

• Server-side auth: Use JWT tokens generated with API secret for server-side operations
• Query endpoints are POST: Channel and user queries use POST with filter conditions
• Filter syntax: Uses MongoDB-style query operators ($gt, $lt, $eq, etc.)
• Rate limits: 60 requests/minute for most endpoints
• Pagination: Use limit and offset; max 100 per request