Shay
955fec3bc3
feat(kernel): route proportional-decrease through proposal-first seam ( #841 )
2026-06-20 15:37:40 -07:00
Shay
168c9f26ac
feat(kernel): add foundational family registry schema ( #840 )
2026-06-20 15:01:27 -07:00
Shay
5cd377b1b8
docs(substrate): add foundational family specification appendices ( #839 )
...
* docs(substrate): add foundational family specification appendices
* docs(substrate): clarify family spec capability wording
* chore: retrigger required checks
2026-06-20 13:49:29 -07:00
Shay
f4a2a2925b
docs(decisions): propose ADR-0224 foundational substrate readiness map ( #838 )
2026-06-20 12:34:31 -07:00
Shay
fbccb2a298
feat(kernel): route proportional-decrease through construction proposals ( #837 )
...
* feat(kernel): route proportional-decrease through construction proposals
* docs: clarify PR #837 scope in lookback
2026-06-20 11:52:02 -07:00
Shay
ec579339ea
feat(kernel): introduce construction-affordance catalog skeleton ( #836 )
2026-06-20 11:32:43 -07:00
Shay
0a168bfb99
feat(kernel): implement ProblemFrame proportional-change closure (PR slice 1) ( #835 )
2026-06-20 10:58:43 -07:00
Shay
b8330477a4
chore: remove private outreach dossier from public repo
2026-06-20 10:21:58 -07:00
Shay
4a1a457655
chore: remove private outreach dossier from public repo
2026-06-20 10:21:51 -07:00
Shay
a6fb58eb46
docs: plan substrate-mediated problem solving ( #834 )
2026-06-20 10:19:55 -07:00
Shay
60df4fc0f9
docs: audit semantic substrate affordances ( #833 )
2026-06-20 09:56:47 -07:00
Shay
564e360eb1
docs(adr): require semantic substrate affordance audit
2026-06-19 18:41:06 -07:00
Shay
6d827eb433
docs(session): record substrate affordance foundation pivot
2026-06-19 18:40:28 -07:00
Shay
9fcc029a13
feat(kernel): bind ProblemFrame relations and measure contract readiness ( #831 )
2026-06-18 21:05:37 -07:00
Shay
956523fd1d
feat(kernel): operationalize ProblemFrame and deprecate legacy parsing ( #830 )
...
* feat(kernel): operationalize ProblemFrame and deprecate legacy parsing
Make #829 kernel substrate the preferred construction path via
build_problem_frame, legacy parsing audit, no-new-legacy agent rules,
morphology planner v2, and guard tests. No serving score or report changes.
* fix(kernel): require actionable morphology planner targets
* fix(kernel): clean ProblemFrame builder diagnostics
* fix(kernel): make morphology planner targets actionable
2026-06-18 20:02:28 -07:00
Shay
95350276e1
docs(kernel): add operationalization deprecation brief
2026-06-18 18:47:38 -07:00
Shay
58a94c8e4b
feat: implement kernel substrate tranche 1 base-layer foundations ( #829 )
...
* feat: implement kernel substrate tranche 1 base-layer foundations
* fix(kernel): ground scalar spans and tighten morphology labels
Add extract_scalar_candidates() with exact source spans and problem_text
provenance while keeping canonicalize_scalar() as the detached pack helper.
Morphology labels now emit missing_* only when substrate frame/unit lookups
actually fail, not on mere trigger-surface presence.
2026-06-18 18:46:20 -07:00
Shay
2a984b86c8
docs(kernel): add broad substrate tranche brief
2026-06-18 17:18:28 -07:00
Shay
51498ca434
docs(kernel): add scalar equivalence PR1 handoff
2026-06-18 17:08:09 -07:00
Shay
035a01f44b
docs(architecture): define kernel knowledge layer inventory ( #828 )
2026-06-18 17:02:18 -07:00
Shay
57a62043d1
feat(gsm8k): xhigh capability sprint 13 robust lift ( #827 )
...
* test(gsm8k): pin xhigh sprint13 confusers and completeness
Add the Sprint 13 contract bundle with target-chain pins, sealed-wrong
neighbors, fraction-word refusals, and non-vacuous completeness cases for
extra mile and extra-actor obligations.
* feat(gsm8k): add xhigh sprint13 verified organs
Gate A2t bounded_rate_projection admits 0016 and 0034 with honest
affine/percent derivations. Gate A2u closed_reference_affine_aggregate
admits 0027 and 0039 with statement-scoped numeric obligations and
repaired comparative provenance. Serving lift: 26/24/0 -> 30/20/0 with
wrong=0 preserved.
* docs(gsm8k): add xhigh sprint13 lookback
Record XHIGH draft recovery, unsafe evidence findings, repairs, Path A
salvage decision, validation outputs, and Sprint 14 recommendation.
2026-06-18 16:10:15 -07:00
Shay
69febc6c0e
feat(workbench): reconcile capability mastery surfaces ( #826 )
...
* feat(workbench): reconcile capability mastery surfaces onto main
Reconcile draft PR 821 onto current origin/main after PR 824 without
porting stale queries/types deletions or fake live API wiring.
- Add Experience Flywheel panel with honest empty state and CLI guidance
- Add Capability Paradigm panel with documented gate ladder A2e–A2q
- Integrate CapabilityMasterySection into EvalsRoute landing and lane view
- Preserve ADR-0160/0162 evidence-first posture; no mutation endpoints
- Add capability-mastery-implementation-plan and panel tests
Invariant protected: wrong_total == 0 serving honesty — documented
24/26/0 from lookbacks, never fabricated as live workbench metrics.
* fix(workbench): keep eval landing guidance above mastery panels
* fix(workbench): document capability milestones through sprint12
* fix(workbench): render capability ladder through sprint12
* test(workbench): assert sprint12 capability milestones
* docs(workbench): update capability mastery plan through sprint12
* fix(workbench): make flywheel empty-state command runnable
* test(workbench): expect runnable flywheel command
* docs(workbench): use runnable flywheel command
2026-06-18 15:40:01 -07:00
Shay
6daa5dfdea
feat(gsm8k): capability paradigm sprint 12 singleton contract lift ( #825 )
...
* feat(gsm8k): capability paradigm sprint 12 singleton contract lift
Add Gate A2r nested_fraction_remainder_total (0004) and Gate A2s
loose_crayon_box_capacity (0007) under explicit ClusterContracts with
hard 0026/0047 neighbor confuser refusal. Serving lifts 24/26/0 to
26/24/0 with wrong=0 preserved; report.json and sealed artifacts
untouched.
* fix(gsm8k): bind nested fraction outer camp
* fix(gsm8k): bind loose box actor and item units
* test(gsm8k): cover sprint12 binding confusers
2026-06-18 15:10:10 -07:00
Shay
50e8de03fe
feat(gsm8k): capability paradigm sprint 11 cluster-contract lift ( #824 )
2026-06-18 12:52:32 -07:00
Shay
a37f390009
feat(derivation): capability paradigm sprint 10 frontier lift ( #823 )
...
* feat(derivation): capability paradigm sprint 10 frontier lift
Add Gate A2o affine_comparative_inversion_total (0009) and Gate A2p
sequential_comparative_scale (0006). Reject wholesale multiplicative_aggregate
and defer 0013 piecewise calendar until month day-count grounds in text.
Serving: 21/29/0 → 23/27/0, wrong=0 preserved. report.json and sealed
artifacts untouched.
* fix(gsm8k): bind affine inversion total question subject
* fix(gsm8k): bind sequential scale page question subject
* test(gsm8k): cover sprint10 subject binding
* fix(gsm8k): license affine inversion aggregate units
* fix(gsm8k): bind sequential scale factors to reading chain
* test(gsm8k): cover sprint10 aggregate and scale confusers
2026-06-18 12:15:52 -07:00
Shay
80753e13de
fix(gsm8k): harden sprint 9 temporal and affine gates ( #822 )
...
Patch post-merge wrong=0 hazards from #820 review without reverting
Sprint 9 organs: bind affine/temporal answers to asked subjects and body
actors, anchor overtime thresholds to per-shift/day surfaces, fix affine
completeness double-counting, narrow percent_partition group skip, and
expand temporal question/rental verb recognition.
2026-06-18 10:57:36 -07:00
Shay
63a9bea823
feat(derivation): capability paradigm sprint 9 temporal tariff + affine lift ( #820 )
...
Gate A2m promotes narrow temporal_tariff organs for overtime shift earnings
(0001) and bundle-overflow rental tariffs (0017). Gate A2n promotes affine
fraction-delta chains for 0010-class references (reference × N/M + K).
Serving moves 18/32/0 → 21/29/0 with wrong=0 preserved. Hardens
percent_partition to refuse subgroup-as-total confusers surfaced in review.
2026-06-18 09:59:46 -07:00
Shay
282c5126de
feat(derivation): capability paradigm sprint 8 R6 affine/fraction lift ( #819 )
2026-06-18 01:52:25 -07:00
Shay
5d529b425f
feat(derivation): capability paradigm sprint 7 experience-guided lift ( #818 )
...
* feat(derivation): capability paradigm sprint 7 experience-guided lift
Add Gate A2i round_trip_trip_duration (0030/cv-0006) and Gate A2j
giveaway_target_residual (0035/cv-0021) as narrow self-verified promotion
bridges discovered via Experience Flywheel decomposition of the
joint-refusal cluster.
Serving: 14/36/0 → 16/34/0, wrong=0 preserved. Holdout_dev admits 0.
* fix(gsm8k): bind giveaway residual to owner
* test(gsm8k): cover giveaway owner binding
2026-06-17 22:47:02 -07:00
Shay
e2f3d37373
feat(derivation): capability paradigm sprint 6 experience-guided lift ( #817 )
...
* feat(derivation): capability paradigm sprint 6 experience-guided lift
Experience Flywheel + scout on post-#816 main showed lift_refused_to_correct=0
(Sprint 5 already served prior deltas). Decompose the MA cluster and R5 pin into
two narrow organs:
- Gate A2g duration_segment_total (0015 / cv-0022)
- Gate A2h survey_rate_earnings (0045)
Ephemeral train_sample: 12/38/0 → 14/36/0, wrong=0 preserved.
Holdout_dev: 0 new admissions. report.json and sealed artifacts untouched.
* fix(gsm8k): require compatible duration units
* fix(gsm8k): harden survey earnings verification
* test(gsm8k): cover sprint6 review hardening
2026-06-17 22:22:53 -07:00
Shay
fd3fc656da
Merge pull request #816 from AssetOverflow/feat/gsm8k-experience-flywheel-pr1-bounded-practice-memory
...
feat(gsm8k): add bounded experience flywheel for sealed practice
2026-06-17 21:52:22 -07:00
Shay
ff055d509b
test(gsm8k): cover experience flywheel review hardening
2026-06-17 21:39:44 -07:00
Shay
b112200654
fix(gsm8k): harden experience compaction families
2026-06-17 21:38:17 -07:00
Shay
9e7432748d
fix(gsm8k): patch experience flywheel merge, provenance, and layering
...
- merge_compacted_runs merges key-by-key without re-expanding prior counts
- load default train_sample cases for live operation_class resolution
- hash full scout row evidence in source_report_hash and source_run_id
- replace scripts.gsm8k_frontier_report import with local _extract_category
2026-06-17 21:25:36 -07:00
Shay
0ca48cc9a3
feat(gsm8k): add bounded experience flywheel for sealed practice
...
Introduce deterministic practice-memory infrastructure that adapts sealed
scout output into compact, retention-gated ExperienceRecords with family,
hazard, and promotion-candidate summaries. No serving, corpus, pack, or
report.json mutation — measurement-only adapter for future sprint reuse.
2026-06-17 21:05:02 -07:00
Shay
aa6173af58
feat(derivation): capability paradigm sprint 5 question-bound product lift ( #815 )
...
Practice/scout showed 0003/0021 refuse on DCS no-injection while sealed
resolve_pooled commits correctly. Build Gate A2f typed revenue/weight product
organs narrower than disabled product_bridge; lifts train_sample 10/40/0 to
12/38/0 with wrong=0 and zero holdout_dev admissions.
2026-06-17 20:51:17 -07:00
Shay
e8f5a347d5
feat(derivation): capability strike batch 4 goal-residual lift ( #814 )
...
* feat(derivation): capability strike batch 4 goal-residual lift
Scout-guided Batch 4 re-wires resolve_promotable_goal_residual serving
promotion (Gate A2e) for train_sample 0037. product_bridge stays disabled
per held-out wrong=0 bisection.
Ephemeral train_sample: 9/41/0 → 10/40/0, wrong=0 preserved.
report.json untouched.
* test(gsm8k): refresh composition corpus prose for A2e
* fix(derivation): refuse comparative goal-residual questions
* test(derivation): guard goal-residual against comparative more-than
2026-06-17 20:21:18 -07:00
Shay
fe4f1223e0
feat(derivation): capability strike batch 3 peer-pick lift ( #813 )
...
* feat(derivation): capability strike batch 3 peer-pick lift
Gate A2d peer_partition_question closes train_sample 0025 by binding
the conditional friend-count clause to a multiply-by-(1+N) graph step
after WAVE-A multiplicative_aggregate statement ingestion.
Ephemeral train_sample: 8/42/0 → 9/41/0, wrong=0 preserved.
report.json untouched.
* chore(analysis): normalize batch3 lookback EOF
* chore(derivation): normalize peer partition test EOF
* fix(derivation): require integer peer-pick friend counts
2026-06-17 19:56:37 -07:00
Shay
0215c30bbe
feat(evals): add GSM8K sealed attempt scout ( #812 )
...
* feat(evals): add GSM8K sealed attempt scout
Deterministic train_sample dual-scorer (serving vs resolve_pooled) that
classifies refusal families and ranks lift targets for Capability Strike.
Measurement-only: no serving mutation, no report.json writes by default.
* chore(analysis): normalize sealed scout lookback EOF
2026-06-17 19:55:13 -07:00
Shay
708f27a240
feat(derivation): capability strike batch 2 bracelet-yield lift ( #811 )
...
* feat(derivation): capability strike batch 2 bracelet-yield lift
Gate A2c adds container_of_product composition for "N bags of M unit"
acquisition statements and yield_question binding that injects unit_partition
from a conditional per-unit rate clause. Live ephemeral train_sample moves
7/43/0 → 8/42/0 with wrong=0 preserved; case 0008 admitted.
* chore(derivation): normalize A2c test formatting
* chore(analysis): normalize batch2 lookback EOF
* test(derivation): guard A2c pronoun ambiguity
2026-06-17 19:32:42 -07:00
Shay
5a0423cb36
feat(derivation): Gate A2b case 0002 fractional rest composition ( #810 )
...
* feat(derivation): Gate A2b case 0002 fractional rest composition
Add fraction_portion operation for "gives N/M of that" and "half of the
rest" subtract semantics chained after unit_partition, plus keep-on-hand
question binding with partition-noun unit inference. Live train_sample
moves 6/44/0 to 7/43/0 with wrong=0 preserved; confuser-v1-0007 still
refuses without "of that".
* test(derivation): add A2b sibling and confuser anti-overfit cases
Cover Bob/Alice partition+fraction chains, slash-without-referent refusal,
partition-less "of that", and multi-actor pronoun ambiguity.
* chore(derivation): normalize A2b fraction test EOF
2026-06-17 19:08:24 -07:00
Shay
65405f1128
feat(derivation): Gate A2a unit partition injection ( #809 )
...
* feat(derivation): Gate A2a unit partition injection
Add typed unit_partition primitive with PartitionChunk/result_unit
contract, recognizer-injector bridge, DCS yield guard, and pronoun
lookback support. Closes unit_partition recognized_no_injection on live
train_sample (0002 partition stmt reclassifies); wrong=0 preserved.
* test(gsm8k): harden unit partition confusers
* test(gsm8k): add unit partition pronoun safety regressions
* chore(gsm8k): fix unit partition exemplar file ending
* chore(derivation): type unit partition solution step operand
2026-06-17 18:14:24 -07:00
Shay
05c5cfa268
docs(analysis): ratify Gate A2a unit partition primitive ( #808 )
2026-06-17 17:37:31 -07:00
Shay
e134e53cca
docs(tooling): add post-Gate-A1 frontier microscope ( #806 )
...
* docs(gsm8k): post-Gate-A1 frontier microscope and roadmap calibration
Add ephemeral live refusal taxonomy with 44-row case-level table,
blocking-layer/primitive/movement classification, and Gate A2a
ratification-candidate recommendation (not implementation). Update
roadmap v2 and Gate A1 lookback metadata (#805 head c37f97bebb083004#806
Remove exact live top-bucket count assertions per #804 doctrine; keep
structural invariants and case 0002 Gate A2a seed pin. Fix EOF newlines
and duplicate _PRIMITIVE_BY_SUBFAMILY key.
2026-06-17 15:53:29 -07:00
Shay
7e1951915a
fix(ci): re-pin public_demo lane SHA after showcase content drift ( #807 )
...
Gate A1 and related cognition-path changes shifted the showcase
determinism fingerprint (86db25b6…) while the lane pin still reflected
the pre-drift artifact (e323adb3…). Re-pin to the CI-verified SHA
2895df08… and refresh v1_dev.json + CLAIMS.md.
Hermetic demo hygiene: use no_load_state for register-tour, learning-
loop, and multi-hop showcase adapters, and isolate CORE_ENGINE_STATE_DIR
in the public_demo lane runner so local engine_state/ pollution cannot
affect eval output or wall-clock.
2026-06-17 15:42:40 -07:00
Shay
99fb14e1ef
Delete .private/san_diego_partnership_packet.md
2026-06-17 14:27:40 -07:00
Shay
bb0830046a
feat(gsm8k): Gate A1 multiplicative comparative recognizer injection ( #805 )
...
* feat(gsm8k): Gate A1 multiplicative comparative recognizer injection
Add COMPARATIVE_WITH_UNIT matcher/injector emitting compare_multiplicative
for the closed v1 template family (twice/thrice/N-times/half/quarter/third).
DCS yields comparative surfaces instead of detection-only fallback.
Includes ratified exemplar corpus + accepted recognizer proposal, 19 unit
tests, and live frontier proof that comparative_with_unit no-injection = 0.
wrong=0 preserved; no report.json rebaseline.
* fix(gsm8k): tighten Gate A1 N-times factors and confuser tests
Restrict comparative matcher N-times factors to plain digits and
single-word cardinals; refuse money, slash-fraction, hyphenated, and
indefinite surfaces. Strengthen confuser tests to assert empty injection
for any recognizer category; add graph-level refusal checks. Add Gate A1
lookback doc and EOF hygiene fixes.
* docs(analysis): pin Gate A1 lookback head SHA
* chore(gsm8k): fix Gate A1 file endings
2026-06-17 14:05:06 -07:00
Shay
fa7819a2e7
add: SD robotics partnership research packet (.private)
2026-06-17 14:00:38 -07:00
Shay
ed2d04c99e
chore(gsm8k): split score preservation from truth preservation in tests ( #804 )
...
* chore(gsm8k): split score preservation from truth preservation in tests
Exact 6/44/0 pins remain only on the committed report.json historical
fixture. Live serving and practice seal tests use a shared monotonic
contract (wrong==0 hard; correct>=6; refused<=44) so capability lift
does not fight the suite.
* chore(gsm8k): polish monotonic baseline test hygiene
2026-06-17 13:30:18 -07:00
Shay
2dc699d3de
docs(analysis): ratify Gate A1 multiplicative comparative injection ( #803 )
2026-06-17 13:20:32 -07:00
