Skip to content

Make evolution domain-aware and preserve behavioral families - #2

Merged
TheDarkLightX merged 1 commit into
mainfrom
feat/domain-aware-evolution
Oct 24, 2026
Merged

TheDarkLightX merged 1 commit into
mainfrom
feat/domain-aware-evolution

Conversation

@TheDarkLightX

Copy link
Copy Markdown
Owner

EvoFuzz now implements the evolvability design: payment seeds can generate domain-valid identity, authorization, replay, reversal and recovery sequences, and the search retains different families instead of evicting by age alone.

Changes:

  • Optional bounded enum/integer parameter domains, identity rebinding, named action blocks and required setup prefixes.
  • A bounded family archive with quality and structural-diversity representatives; extra slots are reclaimed before a whole family is removed.
  • Per-family/operator offspring-productivity learning and parent scheduling. Uniform exploration bypasses both model weights and queued proposals.
  • Exact-request-bound family steering for the LLM, plus domain/archive/statistics context.
  • Parent case/event lineage with explicit duplicate, splice, random-restart and reduction records.
  • A payment-domain example starting only from pay(A), pay(A), and updated design, protocol, agent guide and validation documentation.
  • Development package version 0.2.0; no release tag or package publication.

Validation at d57b52f:

  • 49 local tests pass, including the original target workflows and new domain, retention, learning, lineage and proposal-starvation controls.
  • Eight fresh CLI journeys pass. The 200-evaluation payment campaign retains all five named families plus exploration, finds the deliberately planted double-payout defect, and replays it.
  • The 0.2.0 wheel matches the tested source and passes installed version/search/replay outside the checkout.
  • Lint, compilation, skill-format, public-source privacy and link checks pass.

Scope: family labels are supplied search descriptors, not semantic-equivalence proofs. Multiple descendants may explore the same behavior; duplicates are logged. Online scheduling is measured reward adaptation, not training the LLM or a benchmark claim of improved bug discovery. Existing oracles and process-isolation limits remain unchanged. The separate resonance-retrieval paper investigation adds no dependency or acceptance gate.

@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Oct 24, 2026 •

Copy link
Copy Markdown

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review ✅ Completed 2026-10-24T17:19:46.087533Z d57b52f PR opened
ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: d57b52f4fc

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Comment thread evofuzz/engine.py
self.used += 1
cfg = self.config
actions = p.case(actions, cfg.vocabulary, cfg.limits["max_actions"])
actions = p.case(actions, cfg.vocabulary, cfg.limits["max_actions"], cfg.domain)

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Badge Preserve the required setup prefix during shrinking

With a nonempty domain.setup and sufficient shrink budget, a confirmed failure aborts the campaign because shrink() still deletes chunks starting at index zero, while this validation now rejects candidates missing the setup prefix. The resulting ValueError escapes before the finding and summary are written, losing the confirmation receipts. Reproduced with the payment example using setup [restart] and seed [restart, pay(A), refund(A)]. Restrict shrink deletions to the suffix after the required setup so both search and explicit minimization can preserve replayable findings.

AGENTS.md reference: AGENTS.md:L8-L9

Useful? React with 👍 / 👎.

@TheDarkLightX
TheDarkLightX merged commit 3386b3d into main Oct 24, 2026
12 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant