Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note3 min read

Proof-or-Stop: Don't Trust the Agent, Trust the Evidence -- Loop Engineering for Verifiable Evidence-Gated Lifecycle Control

Source

arXiv

arxiv.org/abs/2607.14890v1

Read the source

Source

Plain-Language Brief

The paper presents Proof-or-Stop Lifecycle Control, a method that ensures transitions in software development lifecycle only happen when there's verifiable evidence to support it. The method was tested through various scenarios and showed improvements in reducing false claims and amplification of issues compared to a naive loop.

What the Document Says

  • The paper introduces Proof-or-Stop Lifecycle Control, a method that allows lifecycle transitions only when fresh, tracked evidence satisfies the relevant gate.
  • The method treats agent outputs as claims and uses proof operationally to mean gate-admissible evidence under a stated trust model.
  • An open-source implementation was tested through mechanism tests, a controlled policy ablation, and self-application evidence.
  • In 10 out of 10 scenarios, the unattended-loop engine passed with zero false DONE claims.
  • Local-key receipt bundles rejected 18 tamper classes with zero false accepts.
  • In a 9,240-cell ablation, Proof-or-Stop reduced visible-pass/hidden-fail amplification from 31 of 1,800 injected cells to 2 of 1,800, showing a 1.6 percentage-point improvement in not-amplified rate.
  • The self-application corpus contains 565 stories and 1,007 review findings with 94.8% resolution.
  • The results suggest that Proof-or-Stop can be a model-agnostic, host-neutral control layer for deciding which claims a lifecycle may act on.

Key Concepts Mentioned

  • Proof-or-Stop Lifecycle Control
  • autonomous coding agents
  • software development lifecycle
  • verifiable evidence
  • gate admissible evidence

Claims Made by the Document

  • Proof-or-Stop Lifecycle Control can effectively permit lifecycle transitions based on verifiable evidence.
  • The method can reduce false claims and amplification of issues compared to a naive loop.

Evidence Used

  • Successful passing in 10 out of 10 scenarios with zero false DONE.
  • Rejection of 18 tamper classes with zero false accepts.
  • Reduction in visible-pass/hidden-fail amplification from 31 of 1,800 to 2 of 1,800 in a 9,240-cell ablation.
  • Resolution of 94.8% of the self-application corpus.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Autonomous coding agents increasingly execute multi-step software work, but lifecycle states such as reviewed, tested, DONE, and ready-to-merge remain claims unless supported by current evidence. We present Proof-or-Stop Lifecycle Control, a method that permits lifecycle transitions only when fresh, tracked-source-state-bound, mechanically verifiable evidence satisfies the relevant gate. The method treats agent outputs as claims rather than lifecycle state, and uses proof operationally to mean gate-admissible evidence under a stated trust model, not semantic program correctness. We evaluate an open-source implementation through mechanism tests, a powered control-policy ablation, and operated self-application evidence. The unattended-loop engine passed 10 of 10 scenarios with zero false-DONE, and local-key receipt bundles rejected 18 tamper classes with zero false accepts. In a 9,240-cell ablation, the pre-registered A4 versus A2-prime comparison reduced visible-pass/hidden-fail amplification from 31 of 1,800 injected cells under a compute-budgeted naive loop to 2 of 1,800 under the gated loop, a 1.6 percentage-point improvement in not-amplified rate with a 95 percent confidence interval of [0.8, 2.5]. A near-compute A3 versus A4 comparison, 14 of 1,800 versus 2 of 1,800, indicates that the gain is associated with enforcing review as a lifecycle gate rather than merely adding a reviewer. The self-application corpus contains 565 stories and 1,007 review findings, with 94.8 percent resolved, plus a 68-row high/critical cross-vendor exhibit. These results support Proof-or-Stop as a model-agnostic, host-neutral control layer for deciding which autonomous-agent claims a lifecycle may act on. The evaluation is limited to one model family, 24 ablation tasks, and a self-hosted corpus.