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

Researchers propose Proof-or-Stop Lifecycle Control, a method that ensures software work transitions only when supported by verifiable evidence. This method improves the reliability of autonomous coding agents by treating agent outputs as claims and using proof to mean admissible evidence under a stated trust model, rather than semantic program correctness.

What the Document Says

  • The paper presents Proof-or-Stop Lifecycle Control, a method for permitting lifecycle transitions only when fresh, tracked-source-state-bound, mechanically verifiable evidence satisfies the relevant gate.
  • The method treats agent outputs as claims and uses proof to mean gate-admissible evidence under a stated trust model, not semantic program correctness.
  • An open-source implementation was evaluated 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.
  • The self-application corpus contains 565 stories and 1,007 review findings, with 94.8 percent resolved.

Key Concepts Mentioned

  • autonomous coding agents
  • lifecycle control
  • proof-or-stop method
  • evidence
  • verifiable evidence

Claims Made by the Document

  • Proof-or-Stop Lifecycle Control reduces visible-pass/hidden-fail amplification compared to a compute-budgeted naive loop.
  • The Proof-or-Stop method improves the reliability of autonomous coding agents.

Evidence Used

  • 10 out of 10 scenarios passed with zero false-DONE in mechanism tests.
  • Local-key receipt bundles rejected 18 tamper classes with zero false accepts.
  • A pre-registered A4 versus A2-prime comparison showed a reduction in visible-pass/hidden-fail amplification in a 9,240-cell ablation.
  • The self-application corpus contained 565 stories and 1,007 review findings with 94.8 percent resolved.

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.