RELAPSE

Security patches are frequently incomplete. A fix guards one code path against a defect — a missing NULL, bounds, or status check — but misses a sibling path with the same structure, corrected in a later commit. RELAPSE asks one narrow, falsifiable question: can a purely structural signal separate genuine missed siblings from look-alikes better than trivial baselines?

It does not claim to have built such a detector. It contributes the apparatus to test the claim rigorously — a frozen detector, a sealed provenance-tracked benchmark, documented evaluation pitfalls with a protocol that avoids them, and an exploratory replication — while staying explicit about what it can show today (engineering soundness) versus what it is built to determine later (a confirmatory result under independent review).

For a function f with multiple call sites, a fix at commit c0 introduces a guard — a conditional on a nearby API call, a NULL/bounds/status check — at some call sites but not at a structurally similar sibling. A later commit c1 adds the analogous guard at the sibling. The episode is the pair (c0, c1) together with the sibling call site and the guard signature.

What it looks like (illustrative sketch — not a corpus episode):

/* commit c0 - the incomplete fix: one call site gets a guard */
buf = read_chunk(src_a);
+ if (buf == NULL) return ERR; /* guard added here... */
process(buf);
...
buf2 = read_chunk(src_b); /* ...but this sibling call site */
process(buf2); /* is left unguarded */
/* commit c1 - later: someone notices and fixes the missed sibling */
buf2 = read_chunk(src_b);
+ if (buf2 == NULL) return ERR;
process(buf2);

The episode is the pair (c0, c1). The detector's job — the open question — is to look at c0 alone and rank the unguarded sibling above the thousands of look-alike call sites that are fine.

Everything needed to reproduce, extend, or contest the benchmark ships together — nothing depends on a live service or a mutable dataset.

A method is only credible if it beats the boring explanations. The frozen ranker must outrank all three trivial baselines, with uncertainty attached to every number.

These are general to structural-signal evaluations, and the protocol is reusable beyond this detector. Each was measured, not hypothesized.

tar xzf missed_sibling_framework_v1.0.tar.gz && cd missed_sibling_framework_v1.0
python3 evaluation/run_benchmark.py fixtures/heldout_SYNTHETIC_DRYRUN.csv "demo"
# expect: [genuine] base=0.480 | AP ranker=0.472 CI(0.32,0.66) | specificity=0.442 | family=0.476

764e122047b580ce34ddb2d153bcdc0c2da0f2aaad172b8fef6aac4bfb856f3b README.md
685bb0941f6b880fd7ce98bc9a85da4fef27ca36b5b0b8ee95d2d1a1f7984aa5 LICENSE
828065bf2e0cd69fede094a63151613f39c4dfe44aa8e4d1e1b852bd2fd0e949 CITATION.cff
4416a80ff785cebad54a97b4c2c279f42863990dcebcace44a0a30cbb566b279 CHANGELOG.md
7ad21a3b1994463647993dbc9a50bf307ebefb85473f1bdd7adbdc45704d72ab PAPER_DRAFT.md
9bea79e1150d16189b982dbbfa0bc03e944f2f57896f208653cd19630a70e654 ARTIFACT_FROZEN_v1.0.md
df80dcc4b6bb0c92dd7dd369391ed3a9051e67fd5740749decf8fb3958ab9fa6 REPRODUCIBILITY.md

Python 3 standard library only for the evaluator and runner. The synthetic regression fixture reproduces the pipeline end-to-end — no corpus clone, no labels.

# 1 · one-minute pipeline check (SYNTHETIC labels — pipeline only, NOT evidence)
python3 evaluation/run_benchmark.py fixtures/heldout_SYNTHETIC_DRYRUN.csv "oracle_v1.0(SYNTHETIC)"
# expect: [genuine] base=0.480 | AP ranker=0.472 CI(0.32,0.66) | specificity=0.442 | family=0.476

# 2 · Tier A — benchmark your own detector (no new labels)
#     features: event_id,signature,named_api,guard_jaccard,family_size,n_uncovered,generic
python3 evaluation/run_benchmark.py <labels.csv> "<your-method>" <events.json>

# 3 · Tier B — confirmatory accuracy (bring two INDEPENDENT reviewers)
python3 labeling_kit/resolve_labels.py reviewerA.csv reviewerB.csv --run
#   agreements auto-resolve; any disagreement blocks COMPLETED until adjudicated

Load the synthetic demo, upload a CSV, or paste your own labeled CSV, then Run.