BugWolf v1.2.11 — Carlini Loop track

v1.2.11 — Carlini Loop track: per-file brute-force discovery

Applies the 2026 per-file brute-force zero-day discovery pattern (Carlini Loop / nano-analyzer / NOVA) to a local project, wired into the existing ZeroDayResearchEngine pipeline.

New

• tools/carlini_loop.py: deterministic project walk → per-file security briefing (imports, functions, entry points, line-anchored dangerous sinks from a NOVA-style catalog) → one research unit per file with CTF framing → model-free offline floor scan → harness-finding intake through the zero-day engine (evidence + novelty dedup + optional chain synthesis).
• Idempotent intake: repeated registration never piles up duplicates; near-matches return likely_variant. Candidates stay HYPOTHESIS until trigger+impact evidence exists; nothing is labeled a zero-day without human review.
• 20 new tests (tests/test_carlini_loop.py). Full suite: 940 passing.

Also in this release

• Exploit-feedback hardening, F0.5 gates, self-eval harness, live execution loop (v1.2.10 era work) + full audit map.

Bundles

• bugwolf-v1.2.11.skill — Claude.ai upload (SKILL.md at root)
• bugwolf-v1.2.11.freebuff.zip — Freebuff/Codebuff (.agents/skills/bugwolf/… layout; install with npx skills add youseefhamdi/bugwolf --skill bugwolf)

v1.2.0: Get zero days

Upgrade: expanded hardening regression tests and kill-chain/patch-gap…

Bugwolf v1.1.0

Upgrade: expanded hardening regression tests and kill-chain/patch-gap…

BugWolf v1.0.0

First public release. Includes the .skill bundle for Claude.ai web/app upload (Customize → Skills).