Summary / The 27 components
05 Issue and fault localization
Turning "feature X is broken" into a ranked list of likely files, functions and lines.
Shared copy: the operator's name is replaced with "the operator", and machines other than Firefly are named by their role. Nothing else is changed.
What is built now
A LOCALIZE state exists in the driver's graph, as a pass-through stub.
What is missing
All five channels the spec names (semantic, graph distance, stack frame, spectrum-based (SBFL), history), the rank fusion and agreement test, and the optional rerank call.
Reviewer Claude, 2026-10-04; close-out verdicts 2026-10-23
Reviewer's read: Closed 10-22: built as the hx localize stage; no detectable gain
This is the first real component on the agreed plan after E0 and E3: SBFL (rank code by how often the failing test covers it) put into the LOCALIZE stub. It is the most "non-LLM" of the components, which fits the programme's design rule, and SWE-bench tasks come with a failing test to drive it. The spec's cost worry stands. Full Ochiai needs passing-suite coverage, and running whole suites under trace on Firefly is slow, so Round 1 ranks by failing-test coverage only.
Close-out, 10-22 (pre-registered, boxed, Claude): built for real as stage S1 (localize) of the close-out bundle arm hx (A12): h0b + localize pre-pass + repro-first with failure digest + repro-triggered reflect/repair (2 rounds) + critic with one revision. hx against h0b, verified-hard 38 + pro-hard 20, 3 seeds: 103 against 94 cells (verified 65/114 against 55, pro 38/60 against 39). Gain +3.0 tasks of 58, paired bootstrap CI [−1.0, +7.0]; 4 regressions on tasks h0b always solved (limit 2); 2.9× h0b cost per cell. No detectable gain at this model. The pre-registered rule closes every bundled component together, with no halving round. Terminal: not re-run.
How it is measured no detectable gain (hx, 10-22)
Arm h3 (top-k candidates injected into REPAIR) against h0-v41; the localization itself is scored separately as file and entity recall@5 against the gold patch.
Adoption rules shared by every arm: How the tests work.
Research & attempts
Score history
No scores yet. The first entry is the baseline for suite v1.
Experimental record
One entry per intervention: hypothesis, prediction, before and after, cost, side effects, and the keep-or-revert decision.
No interventions recorded yet.