Runs / P1 / pz-63-agent-capture-batch-off

1 of 1 declared stages · closed: by exit · recorded in the format stage-receipt-chain/0.5 · ran · received by this console

  • Records agree Yes
  • Ran as approved Not declared
  • Signed by a known recorder Yes
  • Committed by a known time Not anchored

Records agree: this console's check held, with the run's output files required. Every check behind this is on Proof

The case

Run
pz-63-agent-capture-batch-off
ran
→
What it was for
agent_step
This answer's trail was checked on — each check that ran held.

Held under the checks that ran; not anchored, so it can't be shown to be the record that was made.

Repeat the check

Run it from the run's folder:

onetrace-verify --trust trust.json --nested-run 1 MANIFEST.json

What the evidence establishes

Each rung is a separate question, with its own evidence.

Records agreeYes · 82 checks made — all hold.
Signed by a known recorderYes · demo key (public seed: proves the record wasn't changed since signing, not who signed it) ·
sha256:6e20550ae0dc…sha256:6e20550ae0dc6b4471d54336d09d74d14f3f623b735045eddd053c444ec3b14c
Nothing declared is missingYes · Each declared step left a receipt, as the receipts themselves state.
Ran as approvedNot declared · no approved plan recorded
Re-runnable stages reproducedNot checked · No re-check from the record was run, so this page does not say whether anything comes out the same.
Action authorized, effect confirmedNo records · This console does not record authorization.
Committed by a known timeNot anchored · no anchor beside the chain

None of these says the run's output or action was correct.

The execution graph

These verdicts come from the latest comparison of this run against the baseline pz-59-agent-capture-baseline, stored . Open that comparison

agent_step · First difference · 4 output files · 2 steps insideagent_stepFirst difference4 output files2 steps insideinput run:pz-60-agent…input run:pz-60-agent-capture-baseline-subagent differs
  1. agent_stepFirst difference4 output files2 steps insideinput run:pz-60-agent-capture-baseline-subagent differsinput run:pz-64-agent-capture-batch-off-subagent differsoutput model_call.steps.json differsoutput model_call.json differs
Show as table
NodeStageRecorded position
agent_stepagent_step1
FromTo

A link shows which recorded output a stage read. It does not by itself prove that one stage caused the next.

Recorded order

What happened, in the order the receipts give

Held under the checks that ran; not anchored, so none of these steps can be shown to be the record that was made.

StepWhenWhat was checked and what was notOutcome
agent_step
step 1 of 1
→

Checked — it holds.

Completed.
nested run pz-64-agent-capture-batch-off-subagent, its head recorded as
sha256:0e9fa2ca02e2…sha256:0e9fa2ca02e204db4d5a9efee475daf4890ed91a9dfaea4f838ced711e4edd21
· open its page
Receipt · step 1 of 1 · agent_step
step 1 of 1

agent_step

What went in
total · 5 bytes
run:pz-64-agent-capture-batch-off-subagent · 0 bytes
What came out
model_call.steps.json · 684 bytes
model_call.json · 7692 bytes · the model-call record, on its step's page
model_call.2.json · 4020 bytes · the model-call record, on its step's page
return.json · 17 bytes
When
→

Completed.

Ran as agent-planner, version 1.0.0.

Receipt · step 1 of 1 · agent_step
Version
agent-planner · 1.0.0
What went in
total · 5 bytes
sha256:d77ccb69e3f9…sha256:d77ccb69e3f915ed75b4a2db3ff5893fdd0f433e02fdff100441fb843aae585b
run:pz-64-agent-capture-batch-off-subagent · 0 bytes
sha256:0e9fa2ca02e2…sha256:0e9fa2ca02e204db4d5a9efee475daf4890ed91a9dfaea4f838ced711e4edd21
What came out
model_call.steps.json · 684 bytes
sha256:b891d1cf41d0…sha256:b891d1cf41d086267ab96ca798215aeb800c8785fe8a6b68872568d8a026c37a
model_call.json · 7692 bytes · the model-call record, on its step's page
sha256:ff7afbcb8fa6…sha256:ff7afbcb8fa65b123b2ee602f03144c810430f56bdd0d482025ae7e0e7a9aadc
model_call.2.json · 4020 bytes · the model-call record, on its step's page
sha256:97c5212188b9…sha256:97c5212188b9b0ae0cc275225cef3f3ce635cc9f892006471c049454fb6c79fa
return.json · 17 bytes
sha256:3036e8bf5149…sha256:3036e8bf514997c0c20570671c14c5d05fe6a431e1e1ad9b2d931a62df71aabc
Fingerprint
sha256:b77a0016ac05…sha256:b77a0016ac057fae25ccb10d9e41bb56b121674f8d26af7dbbde011008bce0c5
Settings it ran with
{
 "onetrace.kept_inputs": "total"
}
The receipt, exactly as recorded
{"anchor":{"reason":"a demo record; anchoring is not part of this demo","state":"unanchored"},"assertions":{"constants":{"onetrace.kept_inputs":"total"},"undeclared":["output_trust","trust"]},"children":["intake","settle"],"coverage":{"boundaries":[],"completeness":"incomplete","declared_stages":["agent_step","onetrace.close"],"emitting_stages":["agent_step"]},"digest_source":"emitter","emission":{"gaps":[],"policy":"fail-closed"},"format":"stage-receipt/0.5","inputs":[{"bytes":"5","digest":"sha256:d77ccb69e3f915ed75b4a2db3ff5893fdd0f433e02fdff100441fb843aae585b","media_type":"application/json","name":"total","trust_class":"externally-sourced"},{"bytes":"0","digest":"sha256:0e9fa2ca02e204db4d5a9efee475daf4890ed91a9dfaea4f838ced711e4edd21","media_type":"application/json","name":"run:pz-64-agent-capture-batch-off-subagent","trust_class":"operator-authored"}],"instrument":{"config_digest":"sha256:0e5774cac5af774b0eafa55f39a9997651cd54780b1b8c8d85172eda67d60fac","id":"agent-planner","kind":"agent","manifest_digest":"sha256:0f738f99f32591404f16b59a9e8452a0ec71e5ef7f674548103660c268f902a2","rederivable":"true","version":"1.0.0"},"outcome":{"class":"ok"},"outputs":[{"bytes":"684","digest":"sha256:b891d1cf41d086267ab96ca798215aeb800c8785fe8a6b68872568d8a026c37a","media_type":"application/json","name":"model_call.steps.json","trust_class":"model-generated"},{"bytes":"7692","digest":"sha256:ff7afbcb8fa65b123b2ee602f03144c810430f56bdd0d482025ae7e0e7a9aadc","media_type":"application/json","name":"model_call.json","trust_class":"operator-authored"},{"bytes":"4020","digest":"sha256:97c5212188b9b0ae0cc275225cef3f3ce635cc9f892006471c049454fb6c79fa","media_type":"application/json","name":"model_call.2.json","trust_class":"operator-authored"},{"bytes":"17","digest":"sha256:3036e8bf514997c0c20570671c14c5d05fe6a431e1e1ad9b2d931a62df71aabc","media_type":"application/json","name":"return.json","trust_class":"externally-sourced"}],"prev":null,"run_id":"pz-63-agent-capture-batch-off","stage":{"index":"1","name":"agent_step"},"time":{"ended":"2026-11-16T10:14:17.302+00:00","started":"2026-11-16T10:14:17.285+00:00"},"trace":{"span_id":"5117cd53448196d8","trace_id":"4b6460f496052310845afb3282bf95b9"}}
The check's own words
[PASS   ] receipts/01-agent-step.json: receipt format
[PASS   ] receipts/01-agent-step.json: bytes are canonical
[PASS   ] receipts/01-agent-step.json: required field 'format'
[PASS   ] receipts/01-agent-step.json: required field 'run_id'
[PASS   ] receipts/01-agent-step.json: required field 'stage'
[PASS   ] receipts/01-agent-step.json: required field 'prev'
[PASS   ] receipts/01-agent-step.json: required field 'time'
[PASS   ] receipts/01-agent-step.json: required field 'coverage'
[PASS   ] receipts/01-agent-step.json: required field 'emission'
[PASS   ] receipts/01-agent-step.json: required field 'anchor'
[PASS   ] receipts/01-agent-step.json: required field 'instrument'
[PASS   ] receipts/01-agent-step.json: required field 'inputs'
[PASS   ] receipts/01-agent-step.json: required field 'outputs'
[PASS   ] receipts/01-agent-step.json: required field 'assertions'
[PASS   ] receipts/01-agent-step.json: required field 'outcome'
[PASS   ] receipts/01-agent-step.json: time.started carries its zone
[PASS   ] receipts/01-agent-step.json: time.ended carries its zone
[PASS   ] receipts/01-agent-step.json: instrument is named and pinned
[PASS   ] receipts/01-agent-step.json: inputs/total declares a trust class
[PASS   ] receipts/01-agent-step.json: inputs/run:pz-64-agent-capture-batch-off-subagent declares a trust class
[PASS   ] receipts/01-agent-step.json: outputs/model_call.steps.json declares a trust class
[PASS   ] receipts/01-agent-step.json: outputs/model_call.json declares a trust class
[PASS   ] receipts/01-agent-step.json: outputs/model_call.2.json declares a trust class
[PASS   ] receipts/01-agent-step.json: outputs/return.json declares a trust class
[PASS   ] receipts/01-agent-step.json: assertions name their constants
[PASS   ] receipts/01-agent-step.json: outcome class is one of ['error', 'ok', 'refused']
[PASS   ] receipts/01-agent-step.json: coverage declares completeness
[PASS   ] receipts/01-agent-step.json: anchor state is declared, not absent
[NOT-RUN] receipts/01-agent-step.json: originality -- record is declared unanchored; this verifier can establish consistency, not originality
[PASS   ] receipts/01-agent-step.json: digest matches the manifest
[PASS   ] receipts/01-agent-step.json: chain link
[PASS   ] receipts/01-agent-step.json: output artifact 'model_call.steps.json' matches
[PASS   ] receipts/01-agent-step.json: output artifact 'model_call.json' matches
[PASS   ] receipts/01-agent-step.json: record 'model_call.json' parses and is canonical
[PASS   ] receipts/01-agent-step.json: record 'model_call.json' format
[PASS   ] receipts/01-agent-step.json: record 'model_call.json' fields complete and well formed
[PASS   ] receipts/01-agent-step.json: record 'model_call.json' has no value where not observed
[PASS   ] receipts/01-agent-step.json: record 'model_call.json' has no credential material -- by structure only: the verifier is never given the key
[PASS   ] receipts/01-agent-step.json: record 'model_call.json' steps members -- L3, 6 steps
[PASS   ] receipts/01-agent-step.json: record 'model_call.json' steps names an output -- model_call.steps.json, model-generated
[PASS   ] receipts/01-agent-step.json: record 'model_call.json' steps file -- L3: 6 steps, every list lined up
[PASS   ] receipts/01-agent-step.json: record 'model_call.json' model runner -- V1, pinned by the adapter
[PASS   ] receipts/01-agent-step.json: record 'model_call.json' engine core -- in the calling process
[PASS   ] receipts/01-agent-step.json: record 'model_call.json' executor -- uni, as configured and as built
[PASS   ] receipts/01-agent-step.json: record 'model_call.json' L3 sampling -- every member at its covered value
[PASS   ] receipts/01-agent-step.json: record 'model_call.json' how it ended -- not observed
[PASS   ] receipts/01-agent-step.json: output artifact 'model_call.2.json' matches
[PASS   ] receipts/01-agent-step.json: record 'model_call.2.json' parses and is canonical
[PASS   ] receipts/01-agent-step.json: record 'model_call.2.json' format
[PASS   ] receipts/01-agent-step.json: record 'model_call.2.json' fields complete and well formed
[PASS   ] receipts/01-agent-step.json: record 'model_call.2.json' has no value where not observed
[PASS   ] receipts/01-agent-step.json: record 'model_call.2.json' has no credential material -- by structure only: the verifier is never given the key
[PASS   ] receipts/01-agent-step.json: output artifact 'return.json' matches

Where it changed

It changed — here is the step where.

The step
agent_step

Span context

The span context the app reported, as the run recorded it. onetrace does not check it.

The traceparent, trace id and span id are pointers into the app's own tracing tool: they are how to find this run's spans there. This is a public console: it links to no tracing tool.

  • the run's traceparent 00-4b6460f496052310845afb3282bf95b9-5117cd53448196d8-01
  • agent_step
    • trace id 4b6460f496052310845afb3282bf95b9
    • span id 5117cd53448196d8

Steps inside steps

  • agent_step
    • intake · recorded in a nested run
    • settle · recorded in a nested run
    • nested run pz-64-agent-capture-batch-off-subagent, its head recorded as
      sha256:0e9fa2ca02e2…sha256:0e9fa2ca02e204db4d5a9efee475daf4890ed91a9dfaea4f838ced711e4edd21
      · open its page

What changed?

Compared with pz-59-agent-capture-baseline, the most recent earlier run whose final output differs that this demo compared.

What changed?

Other runs with the same inputs: 0

What could be re-checked

No re-check from the record was run, so this page does not say whether anything comes out the same.

This record shows what was verified and evaluated at the time — not what was ultimately true.

Where this answer came from