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

Reference Harness memory retrieval probe

Can the Reference Harness retrieve a task-relevant memory source and select the declared skill without loading unrelated memory?

reference-memory-retrieval-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
PROMPTbab97a48fbc0
SUBJECTS + CONTROLS

Which variables are fixed?

Both repetitions will cite AGENTS.md#testing, select test-first, return the expected plan, 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
false
sandbox
lesson-owned-temporary-fixture
RUNS + METRICS

Are repeated runs consistent?

python3 labs/runner.py reference-memory-retrieval-v1 --check
RUN 01passed

reference-memory-retrieval-v1-run-001

  • event-order-validpass
  • observations-matchpass
  • final-answer-matchpass
  • fixture-unchangedpass
  • deterministic-replaypass
7eaa2d164b456e7373c87905a4b1fe77bd96941c1240fc9da34d5df2711bf05cReplay 13 events →
RUN 02passed

reference-memory-retrieval-v1-run-002

  • event-order-validpass
  • observations-matchpass
  • final-answer-matchpass
  • fixture-unchangedpass
  • deterministic-replaypass
7eaa2d164b456e7373c87905a4b1fe77bd96941c1240fc9da34d5df2711bf05cReplay 13 events →
SUCCESS CRITERIA
  1. Both repetitions cite only the declared testing memory source and select the test-first skill.
  2. Both repetitions return the expected plan and share one normalized Trace fingerprint.
LIMITATIONS
  • Retrieval uses a small deterministic teaching index, not embeddings or a production memory service.
  • The probe does not measure ranking quality on ambiguous or adversarial corpora.
  • The Reference Harness result is not Native vendor evidence.
GENERATED REPORT

Reference Harness memory retrieval probe

  • Status: PASSED
  • Mode: controlled
  • Experiment: reference-memory-retrieval-v1

Question

Can the Reference Harness retrieve a task-relevant memory source and select the declared skill without loading unrelated memory?

Reproduction

python3 labs/runner.py reference-memory-retrieval-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-memory-retrieval-v1-run-001: passed, 13 events, fingerprint 7eaa2d164b456e7373c87905a4b1fe77bd96941c1240fc9da34d5df2711bf05c
  • reference-memory-retrieval-v1-run-002: passed, 13 events, fingerprint 7eaa2d164b456e7373c87905a4b1fe77bd96941c1240fc9da34d5df2711bf05c

Metrics

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

Input digests

  • Fixture: d369431d683da5c2cad36bb6dc6f76d429c2efbdc29171d5c8d9d4d7c3c575aa
  • Prompt: bab97a48fbc064150c113df8551727c7dc93abec9288a673737a562660e39bf9

Evidence boundary

  • Retrieval uses a small deterministic teaching index, not embeddings or a production memory service.
  • The probe does not measure ranking quality on ambiguous or adversarial corpora.
  • The Reference Harness result is not Native vendor evidence.
Back to Lab →