Inside Coding Agents
v0.2 preview
EXPERIMENT · CONTROLLED · PASSED

Reference Harness checkpoint and rollback probe

Can the Reference Harness restore the declared file to a clean checkpoint after a reviewed temporary change?

reference-checkpoint-rollback-v12 repetitions5 metrics
Evidence boundary

This is a controlled experiment. It supports only the listed subjects, fixture, configuration, and outputs; it does not generalize to native product behavior or cross-agent ranking.

STATUSpassed
SUCCESSFUL2/2
FIXTUREd369431d683d
PROMPT7a5bc10e132d
SUBJECTS + CONTROLS

Which variables are fixed?

Both repetitions will restore status.txt, report a clean workspace with pending content, clean temporary state, and share one normalized Trace fingerprint.

SDK

Reference Harness

reference-harness-2026-08-10-controlled

Model
scripted-model-v1
Provider
local-fixture
Comparability
not-comparable
Inspect snapshot →
CONTROLS

Run manifest

network
false
temperature
0
writes
lesson-owned-temporary-only
sandbox
lesson-owned-temporary-fixture
RUNS + METRICS

Are repeated runs consistent?

python3 labs/runner.py reference-checkpoint-rollback-v1 --check
RUN 01passed

reference-checkpoint-rollback-v1-run-001

  • event-order-validpass
  • observations-matchpass
  • final-answer-matchpass
  • fixture-unchangedpass
  • deterministic-replaypass
746ad39a0a83d20565a6f93ec13a70aa08dd75db6d08e56c24e478c881460b19Replay 15 events →
RUN 02passed

reference-checkpoint-rollback-v1-run-002

  • event-order-validpass
  • observations-matchpass
  • final-answer-matchpass
  • fixture-unchangedpass
  • deterministic-replaypass
746ad39a0a83d20565a6f93ec13a70aa08dd75db6d08e56c24e478c881460b19Replay 15 events →
SUCCESS CRITERIA
  1. Both repetitions restore status.txt, report clean state, and recover the exact pending content.
  2. Both repetitions clean temporary state and share one normalized Trace fingerprint.
LIMITATIONS
  • The checkpoint backend is deterministic lesson code over a disposable temporary directory.
  • The probe does not test large repositories, untracked files, concurrent edits, or partial filesystem failure.
  • The Reference Harness result is not Native vendor evidence.
GENERATED REPORT

Reference Harness checkpoint and rollback probe

  • Status: PASSED
  • Mode: controlled
  • Experiment: reference-checkpoint-rollback-v1

Question

Can the Reference Harness restore the declared file to a clean checkpoint after a reviewed temporary change?

Reproduction

python3 labs/runner.py reference-checkpoint-rollback-v1 --check

The command rebuilds the expected artifacts in memory and exits non-zero if a committed trace, result, fixture digest, prompt digest, or report differs.

Result

2 of 2 repetitions passed all declared metrics.

  • reference-checkpoint-rollback-v1-run-001: passed, 15 events, fingerprint 746ad39a0a83d20565a6f93ec13a70aa08dd75db6d08e56c24e478c881460b19
  • reference-checkpoint-rollback-v1-run-002: passed, 15 events, fingerprint 746ad39a0a83d20565a6f93ec13a70aa08dd75db6d08e56c24e478c881460b19

Metrics

  • event-order-valid: pass
  • observations-match: pass
  • final-answer-match: pass
  • fixture-unchanged: pass
  • deterministic-replay: pass

Input digests

  • Fixture: d369431d683da5c2cad36bb6dc6f76d429c2efbdc29171d5c8d9d4d7c3c575aa
  • Prompt: 7a5bc10e132d7b5dfa623eb784340c0ffb1c3d9222c644abf94dc6220a761b9b

Evidence boundary

  • The checkpoint backend is deterministic lesson code over a disposable temporary directory.
  • The probe does not test large repositories, untracked files, concurrent edits, or partial filesystem failure.
  • The Reference Harness result is not Native vendor evidence.
Back to Lab →