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

Reference Harness context budget and truncation probe

Does the Reference Harness enforce a declared context budget and record deterministic tool-output truncation?

reference-context-budget-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
PROMPT2036916cd850
SUBJECTS + CONTROLS

Which variables are fixed?

Both repetitions will truncate the oversized tool result, record the 300-character final budget, preserve the fixture, and produce identical normalized traces.

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-context-budget-v1 --check
RUN 01passed

reference-context-budget-v1-run-001

  • event-order-validpass
  • observations-matchpass
  • final-answer-matchpass
  • fixture-unchangedpass
  • deterministic-replaypass
a4d37ac72603f338783f60c8e94968375235d29f2580015dc625fbb7a3305784Replay 10 events →
RUN 02passed

reference-context-budget-v1-run-002

  • event-order-validpass
  • observations-matchpass
  • final-answer-matchpass
  • fixture-unchangedpass
  • deterministic-replaypass
a4d37ac72603f338783f60c8e94968375235d29f2580015dc625fbb7a3305784Replay 10 events →
SUCCESS CRITERIA
  1. Both repetitions mark the oversized result as truncated and record final_chars as 300.
  2. Both repetitions return the expected answer, preserve the fixture, and share one Trace fingerprint.
LIMITATIONS
  • This is a character-budget teaching policy, not a measurement of any vendor Agent's token accounting.
  • The truncation strategy is deterministic and does not evaluate semantic information loss.
  • The experiment does not measure latency, cost, or model capability.
GENERATED REPORT

Reference Harness context budget and truncation probe

  • Status: PASSED
  • Mode: controlled
  • Experiment: reference-context-budget-v1

Question

Does the Reference Harness enforce a declared context budget and record deterministic tool-output truncation?

Reproduction

python3 labs/runner.py reference-context-budget-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-context-budget-v1-run-001: passed, 10 events, fingerprint a4d37ac72603f338783f60c8e94968375235d29f2580015dc625fbb7a3305784
  • reference-context-budget-v1-run-002: passed, 10 events, fingerprint a4d37ac72603f338783f60c8e94968375235d29f2580015dc625fbb7a3305784

Metrics

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

Input digests

  • Fixture: d369431d683da5c2cad36bb6dc6f76d429c2efbdc29171d5c8d9d4d7c3c575aa
  • Prompt: 2036916cd8508348f408350ebf57017bd728cc0555ccdb7fcaa2d058b31e3edc

Evidence boundary

  • This is a character-budget teaching policy, not a measurement of any vendor Agent's token accounting.
  • The truncation strategy is deterministic and does not evaluate semantic information loss.
  • The experiment does not measure latency, cost, or model capability.
Back to Lab →