Websites & products
Public apps, marketing, platform UIs, and repo tooling
Langfuse
Tracing and project observability
PgAdmin
PostgreSQL admin console
Namecheap
Domain registrar account
Notebooks & analytics
Interactive notebook environments
APIs
HTTP services and gateways
FrontierChats
Open WebUI · chats.finalmoment.ai
Build & delivery
CI/CD servers
Proxy & routing
Reverse proxy and TLS management
Central place for product, platform, and internal roadmap views.
Active Docker containers
Live containers with published ports
Snapshot
All headline metrics in one grid. Larger tiles highlight scale and volume.
Activity
Event mix, active users (DAU / WAU / MAU-style), and top agent owners. Uses the analytics window in the header.
Platform activity (last 30 days)
Active users
Distinct users with chat activity (analytics)
Bar length is relative to the largest of DAU, WAU, and MAU in this snapshot (not a quota target).
Top 10 · Agents
All time · agents owned, one row per email (merged accounts)
User leaderboards
Top accounts by LLM tokens, stored messages, or tool calls. Chart and list stay in sync.
LLM completion tokens (analytics)
Stored session_messages rows (merged accounts)
Tool call events (analytics)
Tool payload tokens (estimate)
Sums use the same window as charts and user leaderboards (chars÷4 on tool args/results).
Top tools by estimated total tokens
Metric explorer
Event timeseries
Daily buckets for the last 30 days (same window as the header control).
Users & activity
Leaderboards for engagement, newest accounts, and a sortable directory.
Top users · user messages
Ranked by user-role turns in chat (30-day style signals).
Newest accounts
Most recently created users, newest first.
Directory
Sort columns to explore agents, sessions, and stored message volume.
Columns
| Status |
|---|
User
Agents & usage
Top cards highlight a few high-volume agents; the catalog below is the full filtered directory (not limited to the cards).
Top agents · stored messages
All rows in session_messages for this agent — strongest signal of real usage.
Catalog
Grouped by owner; sort applies within the filtered result set.
| Owner | Flags |
|---|
Tools & models
Tool reliability, estimated I/O tokens on tool calls, model spend, and recent failures.
Tool call breakdown
Volume, success, latency, and estimated argument/result tokens (chars÷4) attributed to each MCP tool name.
| Tool | Calls | Success | Avg latency | p50 | p95 | Est. in | Est. out | Est. total | Avg in | Avg out |
|---|
tool_call). Chat can work without them if tracing was not enabled historically.
Model usage
LLM completion events: request counts, token totals, and mean latency by model.
| Model | Requests | Input tokens | Output tokens | Total tokens | Avg latency |
|---|
llm_completion events with token counts. If this stays empty, ensure the API logs completions to analytics.
Total tokens by model
Same window as the table — horizontal bars for quick comparison.
Recent tool errors
Failed tool calls from analytics — useful for chasing flaky MCP integrations.
| Time | Tool | Error code | Latency | Agent |
|---|
Base Iris programming
Tune Base Iris (Iris 1.0): pick a base model from Iris GET /v1/models and set reasoning effort. Settings use GET/PATCH /v1/admin/base-iris/settings (read: admin; write: superadmin).
Iris Code
Pin the coding model for the IrisCode agent and set the default reasoning effort for new sessions. Uses GET/PATCH /v1/admin/iris-code/settings (read: admin; write: superadmin). Saves sync the IrisCode agent row.
Invitations
Create invitation
| Role | Status | Expires | Created |
|---|
User suggestions
Free-text comments from the feedback API across all users and agents.
| Updated | User | Agent | Scope | Reaction | Rating | Comment |
|---|