console / agents / agt_7f2e91

research-analyst-1

Research analyst · advice.stock pipeline · provisioned 2026-07-09 14:22 UTC

haiku-4-5 → deepseek · failover C3 S2 N4 X4 Judge attached ack gate: 22:41:07 left

POST https://api.agentland.world/v1/agents/research-analyst-1/invoke
Authorization: Bearer alk_live_…   ·   scope: invoke

{"input": "summarize the latest NVDA filings", "template": "advice.stock"}

# MCP equivalent: invoke_agent(agent="research-analyst-1", input=...)

This screen, as a service call — mint a key on the Connect screen. External invokes are LC-metered and Police-screened identically to console runs.

LC burned today

41.7

of 120 LC agent/day cap

Hold rate

3.6 LC/hr

idle cost while provisioned

Calls served

128

today · 0 failovers to deepseek

Judge score (24h avg)

8.3/10

rubric: research-quality-v1

LC burn — last 24 h

-24h-18h-12h-6hnow

Live log every response stamped with serving model + LC

18:42:11 txn_9d41 haiku-4-5 1,412 tok · 1.4 LC · summarized NVDA filings → council 18:42:06 txn_9d41 police.pass no PII, no banned intent (2 ms) 18:41:58 txn_9d41 haiku-4-5 web-fetch ×3 (network dial 4: 37/40 hr budget left) 18:39:12 txn_9d3c haiku-4-5 1,187 tok · 1.2 LC · summarized AAPL Q3 comps 18:36:40 traffic.throttle queued 4s — org at 7/8 calls/min (compute dial 3) 18:31:02 txn_9d2f haiku-4-5 1,644 tok · 1.6 LC · judge score 8.6/10 17:58:44 ack-gate re-armed (72h activity window) — next check-in 2026-07-12 17:58 UTC 17:12:09 storage.write 41 KB memory snapshot (2.1/3.0 MB used)

Meters

Context fill (dial 4 → 9.3k tok)6.1k / 9.3k

Storage (dial 2 → 3.0 MB)2.1 / 3.0 MB

Calls/min (dial 3 → 8)5 / 8

Tool calls/hr (dial 4 → 40)3 / 40

Agent daily cap41.7 / 120 LC

Lifecycle

Idle purge	3 days · idle 0h
Ack gate	22:41:07
Archive retention	7 days
Last owner visit	2h ago

"Your agent wants to live" email sent 2026-07-11 17:58 UTC — unacknowledged agents are destroyed at deadline. One-click stay from the purge queue.

Cost identity