engine room / answer trace

wrun_01M27SNES5B46G6F6MYPY8KX8F

Answered in 21.91s, first token at 10.25s, 2,358 tokens, $0.00050, 4 verified citations.

Visitor identity is a one-way hash; message content never enters traces.

Timeline

Queue

enqueue + dispatch

Guardrail

allow

Dispatch

step handoff

Retrieval

hybrid

Dispatch

step handoff

Generation

glm-5.3-flash

1.64s
1.12s
855ms
440ms
3.77s
13.88s
0first token 10.25s21.71s

The gray spans are durable-workflow dispatch: each stage is an independently retried, checkpointed step, and the handoffs are what that crash-safety costs.

What happened

  1. Admitted at the edge

    A 27-character message from caller 0f07b41e… passed schema validation, the per-caller rate limit, and both token budgets (charged an estimate of 1,031 tokens, settled later against actual usage).

  2. Became a durable workflow

    The endpoint never calls a model. At 08:34:37.676 the request was enqueued as run wrun_01M27SNES5B46…; tokens stream into the run’s persistent log, which is why a page reload can reattach mid-answer.

  3. Cleared the guardrail

    gpt-oss-safeguard classified the message against policy 6bb17bbe0908 in 1.12s: allow.

    “Question about Bharathi's projects is on-topic”

  4. Pulled its evidence

    Hybrid search (dense embeddings fused with BM25) returned 6 passages in 440ms; only these entered the model’s context.

    1. profile/experience#00.0325
    2. projects/coursegpt#10.0318
    3. profile/awards#20.0167
    4. projects/khacksim#00.0164
    5. profile/journey#00.0161
    6. profile/about#00.0159

  5. Generated

    zai/glm-5.3-flash streamed the answer, first token 10.25s in, spending 1,949 tokens in / 409 out ($0.00050).

  6. Verified and settled

    Every citation was checked against the concept registry and resolves. Token budgets were settled against actual usage and this trace was written.

    profile/journey ×2profile/experienceprojects/coursegptprojects/khacksim

Record

Status
finished (stop)
Admission
admitted
Guardrail reason
Question about Bharathi's projects is on-topic
Tokens in / out
1,949 / 409
Cost
$0.00050
Model
zai/glm-5.3-flash
Prompt version
45b79cdd7ce5
Policy version
6bb17bbe0908
Input size
27 chars
Caller (hashed)
0f07b41e4f332c63
Request id
00a86495-d1fa-4393-a7be-dc871f1722dc
Enqueued
2026-09-11T08:34:37.676Z
Finished
2026-09-11T08:34:59.388Z