Receive structured provider-owned interaction analysis for docking poses (E4)

Start by inspecting the DockingMT provider integration and qualified MolSysMT runtime, then verify the supported assembly and atom mapping for an existing docking pose. Track provider analysis through association, filtering, and persistence while preserving source identities and provenance. Done requires numerical and consumer-association evidence from a bounded representative workflow with positive, adverse, and empty controls; rendering acceptance remains in #52/#60.

Current editable Python interaction-view checkpoint — 2026-11-08

Published a6e63690 ([skip ci]): complete preregistration, outcomes, original-criterion audit and resumption. Open/partial: all eight original Python View/widget/native cells pass; actual browser/comm/canvas stays unreceived under #52/#60.

Execution base 204f5f6c, actual editable MolSysMT fab8cbccf570a60205a3863aad19dd4b7cfb8d3a, Viewer 1187fc187247aa2d20d01791856a541d50562bcb, native SHA-256 5715bda28be19820e96c2b81e4734bf75e1a9f05dd660d9c7e2817bc6bcd79d3. No build/reinstall, provider pull, runtime/helper/data/pin edit or shared-environment modification.

All normal cells precede pm/fs: retained original; explicit five-record permutation selecting collection index3/local frame0; translated evaluated-empty/unevaluated control; unchanged two-frame native widget regression. Original molecular/action blocks are byte-identical. Real Python visuals/widgets, complete result/native/typed equality, queries/export, four refusals per integration cell, foreign ownership, replacement invalidation, cleanup and final source/caller/provider restoration pass. All original scores/state/source identities and four unevaluated poses remain; no detector/search repeats.

351 primary members, manifest 87bf805d6dd1ba2f6ed9075043c1e09f655d1d7d10be9e10e8b0a594905e57ca, and 17 reader members, 3436ee960ffcc0fe6665c8ad52503b1306272b2aa5f0d105f4178cd91096e58e, preserve two zero-operation setup failures and the failed initial reader separately. Final standard-library reading passes in 13.908 s / 719,196,160 sampled RSS bytes, with zero scientific imports/molecular operations. Published 15-file Pytest Receptor1.2.2 custody is separate from the editable runtime and build labels. Every group is terminal, within original caps/CPU0/nice10/one-worker/13GiB reserve, without resource interruption.

Five final workspace gates and four fresh exact committed-export checks pass (indexes, shared resources, publication contract, every primary/reader archive blob); fresh receiving takes 4.202 s / 106,643,456 sampled group RSS bytes. The five original #80 criteria retain their declared one-carrier scope. The maintainer reports uibcdf/molsysmt#256 is being worked on: wait for its supported handoff, preserve E3's lossless-reference/full-workflow gate, and continue independent accepted work without polling/building or a private transport. Public/installed/platform/scientific-panel/full objective and #28 exact-head hosted evidence remain unaccepted; no CI dispatch.

---

Partial-carrier admission correction — 2026-11-08

Published d594fb66 ([skip ci]), archived report and exact evidence. Both public binding routes now detect absent successful atom/frame-count or coordinate returns before coercion/unit extraction and raise DMT-E002 naming molecular_system with the missing domain. Actual provider exceptions propagate unchanged; no data, maps, chemistry, frames or analyses are invented.

Guard: tests/test_interaction_binding.py::test_missing_carrier_domains_have_a_local_argument_diagnostic. All six original regressions fail on the old runtime (four TypeError/two low-level quantity-form errors); that red population remains FAIL. Final complete binding/view source selection passes 78 cases per normal/pm-fs policy, zero failures/errors/skips/deselections, retaining existing valid/negative/coverage/provider-error/detachment semantics. The preformat PASS and two quality-scope/format observations remain separate; formatting preserves all test AST/assertions.

60 primary + nine reader records retain original/final source, raw logs, complete native Pytest Receptor/JUnit, policies and terminal resource outcomes. Supported artifact reading independently verifies agreement without replay. Seven applicable local gates and six fresh committed-export checks pass. Ruff explicitly excludes the untouched untracked migration bundle; the source/export checks pass. One CPU/nice 10/one native worker, monitored 13 GiB reserve; all owned groups terminal without resource interruption.

This resolves #89's local source defect only. No real molecular operation/engine/View, install/build, search/browser/kernel or remote CI occurs. #80/full E4/scientific/installed/canvas and uibcdf/molsysmt#659/#256 E3 holds remain separate. Roadmap, plan, handoff and binding contract are updated; exact-head hosted evidence remains missing under #28.

---

Complete editable catalog-persistence receiving — 2026-11-08

Published c238513a1e76715a56abffb6528ba31bf913e226 [skip ci]: complete receiving and limits, 86-member primary (SHA-256 df0a0f0299d298e282d54c4d8cc62703cc2c69a970876438cb5d8972bb14953c) and eight-member independent-reader companion (11917ca80ad31705aa409f447f42fd3aa7fa98ce594f7f5c785a996e25b26c08).

The original positive-to-unevaluated catalog-persistence selection passes completely under normal and pm/fs on actual editable consumer 542c5c5, MolSysMT 2325038e8 and native SHA-256 5715bda28be19820e96c2b81e4734bf75e1a9f05dd660d9c7e2817bc6bcd79d3. Both the immutable old failing carrier and fresh public binding/full H5MSM roundtrip retain complete analysis/native state, catalog/code identity, original five-pose results and source nonmutation. Hbonds is unevaluated with zero occurrences and its original label; hydrophobic retains two observations. Four other original poses remain unevaluated. All original assertions, tolerances and budgets remain; no detector, preparation, scoring/search or View executes.

A first task-instrumentation failure names obsolete molsysviewer.View before every molecular call; it remains FAIL. Only the forbidden-constructor guard changes to public MolSysView for the complete retry. Normal 24.177 s and pm/fs 24.270 s pass, maximum sampled group RSS 580,562,944 bytes, CPU 0/nice 10/worker 1, monitored 13 GiB reserve, terminal owned groups and no resource interruption. Independent stdlib record reading passes without molecular replay; five final local checks and three fresh committed-export checks pass for 102 changed files. No runtime/helper/data/pin change, build/install or remote CI is launched.

This clears uibcdf/molsysmt#642's consumer persistence hold for the editable one-carrier profile. Original failed runs/private installed artifacts keep their own identities; public/installed/minor/platform and actual canvas gates are separate.

#80 remains partial for its remaining complete workflow/actual canvas criteria. Next receive the original reference selections under #33/#49/#79. uibcdf/molsysmt#604, uibcdf/molsysmt#558 and the complete #73 objective retain their scope.

---

Shared provider artifact admission checkpoint — 2026-11-08

Published 75db240e9c0b3aefa64eae7258c1a513b664f9af [skip ci]: source/registry boundary, 19-member primary records, 17-member independent-reader/publication records.

Exact provider source ad67e0e915ba23d90cb9d6331c063c342852d67a contains the resolved uibcdf/molsysmt#642, uibcdf/molsysmt#644 and uibcdf/molsysmt#646 corrections. Its full Rust tree and source-declared build sections equal uibcdf/molsysmt#642's; the old qualified native has different Rust inputs and is ineligible for this target. All 107 advertised Conda file identities remain unchanged. No covering artifact, installed-client correction or scientific acceptance is received.

Independent record reading, nine local checks and three actual committed-export checks pass for 45 exact changed files. Runtime/helpers/historical scientific inputs remain unchanged; no molecular calls, build, install or remote CI. The original uibcdf/molsysmt#642 construction stays held after its owned-memory stop; old failed selections remain FAIL. Obtain one whole owner-built covering artifact, or an explicitly supported composition with matching qualified native identity/build provenance, then refresh actual routine composition and register the original separate receiving selections. Preserve uibcdf/molsysmt#558, uibcdf/molsysmt#645 and uibcdf/molsysmt#604, canvas/environment, full Core/Extended/scientific/public/stabilization scope and #28's missing exact-head hosted evidence. Task/cache custody remains #50, review 2026-11-14.

---

Acceptance audit and provider-artifact hold — 2026-11-07

Published checkpoint and current resumption route. Open/partial; new-fix client receiving remains unexecuted.

All five original criteria are audited against actual scientific, installed association, coverage and addon evidence. The preceding eight View/native cells and 186 Python/23 DOM cases retain their exact old-provider scope, full original five-pose population and no-broadcast mapping. Actual browser/comm/canvas remains separately held under #52/#60.

MolSysMT #642 is correctly source-resolved at 4d8af6c34cb6b3f9e22833771b246d8767ef85e9. Its owner receives catalog/code fidelity and the already-written failing carrier. DockingMT's separate exact provider construction preserves three nonpassing resource interruptions: two 60 s wall stops, then an owned-group1 GiB RSS stop during faer0.24.4 under a separately registered construction-only time budget. The last sampled peak is 1,075,798,016 bytes, elapsed 202.384 s; available RAM stays above 60 GB. Every group is terminal with no live owned descendant. Do not automatically restart construction after this memory stop. No complete wheel/provider prefix/candidate or scientific receiving cell exists; no old native binary or module is patched/relabelled. Prepared receiving scripts are unexecuted.

44-member interruption archive: f2af7ec1e07cde5883ebd64a973a0ad7367a5ed8c810686008ff17263ed81e25; 19-member administrative-reader companion: 187dd4042d270f8471932e1fc93fab3532e2ef88c01c18a7acd049989ff67164. Independent standard-library byte/classification checks pass, accepting zero construction/scientific cases. All 75 consumer/3794 provider/641 Viewer bytes and routine Conda metadata remain unchanged. Thirteen applicable local checks and three fresh actual-committed-export checks pass on 75 exact checkpoint files. No manual CI; #28 exact-head hosted evidence stays missing under [skip ci]. #50 custody review 2026-11-14; original migration untouched/untracked.

Next for this issue: request a complete owner-built exact-source artifact/native identity or supported acquisition route, then newly admit the candidate and finite positive-invalidation client receiving. Source #642 stays closed; the historical consumer failures remain failed. Independent next work: #83 owns source-free installed reconstruction of the existing #16 1IEP reference/control workflow; audit records/replay/support/tolerances before registering native execution. E3 #604, full Core/Extended #73, independent science and public/minor/platform gates remain unproven.

Earlier snapshots below preserve their original evidence; this disposition supersedes their historical next actions.

---

Installed interaction-view receiving — 2026-11-07

Published checkpoint, registered population/outcomes and resumption route. Open/partial: all eight finite real-provider cells pass; actual browser/comm/canvas, public delivery and whole E4 remain unproven.

• New exact runtime source 2ef82d25e34266d1b504464dc238e8863a9dade0; private wheel 0.0.0+310.g2ef82d2, SHA-256 6db76835f4f3f9c41d06a163cb3a4fbc9e780597bfd89d2fc6a75d29cff31133, 154,750 bytes/75 installed consumer files. Borrowed MolSysMT 9e1f1b1, Viewer 78ede79e overlay, SDK 95f693be and original scientific producer 5952053 retain separate identities. No runtime/test/helper/provider/shared-environment change in this receiving increment; older consumer wheels do not qualify this route.
• Four passing cells per policy, all normal before pm/fs: retained original; explicit five-record permutation selecting original retained pose at index3/position4 while local frame0 and scientific rank/ID stay original; separately identified empty-source native unevaluated/evaluated-empty control; exactly one existing two-frame native widget/callback regression. No new docking, detector, energy, geometry transformation or independent scientific case. All five original records stay exact; the derived control is never a sixth original and the other four poses remain unevaluated.
• Three View cells/policy each render twice in one real View through public dmt.view, create actual provider visuals/two widgets, check live metadata/queries/summary/export, four refusals, owned cleanup/re-enable and equal-count system-replacement invalidation. Source native/typed data survive. Native pytest events/JUnit/exit agree on one pass per policy; cached archive-verified published Pytest Receptor1.2.2 uses a separate 15-file prefix.
• Two receiver setup failures and one label rejection remain preserved. All groups are terminal, resource/call limits and 8 GiB reserve hold, no resource interruption. Independent standard-library saved-record reading passes first attempt in 40.739 s/496,467,968 sampled group-RSS bytes, with zero scientific imports/molecular replay.
• 342-member primary archive: a6b92821b8c5af375fcadbb8e64f6e2cf01015283b95e33ea47af53a3e692606; 17-member reader companion: 50a1acecae7b7ec418b17ed1bef439b49d6660d6c84edc36a75c1ed1c6a44f13. Thirteen applicable local checks and three fresh actual-committed-export checks pass, qualifying 371 exact checkpoint paths. No manual CI; [skip ci] leaves #28 hosted exact-head evidence missing.

Next: audit remaining #80/#60 criteria against this received one-carrier route. Keep the maintainer's browser/comm/canvas #52/#60 scheduling hold. MolSysMT #642 positive-source persistence remains failed with owner triage but no received correction. Other-carrier projection, E3 #604, broader Core/E5, independent scientific reproduction and public/minor/platform delivery retain their own gates. Task custody remains #50/review 2026-11-14; original migration files stay untouched/untracked. Full #73 and the active goal remain unproven.

The source-only and earlier installed snapshots below retain their original artifact/population scope; their historical “next” steps are superseded only by the receiving above.

---

Interaction view implemented and source-qualified — 2026-11-07

Published source, public route/contract, and resumption checkpoint. Open/partial: source and DOM qualification pass; actual new-candidate/provider and canvas receiving remain pending.

• dmt.view(binding) loads the complete admitted one-frame carrier through public MolSysViewer loading/interaction-visual APIs. It retains the entire original result and maps frame 0 only to binding.pose_index, including a nonfirst original member. Explorer, frame callbacks, Python/context/panel navigation and summary/export share that mapping. Other poses remain listed, with unavailable geometry explained and navigation disabled; no analysis broadcasts.
• Live provider metadata distinguishes unevaluated, evaluated-empty and unavailable families. Replacing the public loaded-system handle invalidates the association even with matching counts. Conflicting source/result/pose options fail before view creation/mutation. Cleanup affects owned visuals, retains handles on failures and leaves scientific/foreign data intact during clear/disable; provider scene replacement retains its own semantics. Provider failures propagate and loading is not transactional. No molecular matcher, codec, preparation, detector or geometry projection is added.
• 186 source/protocol Python cases pass, including 28 new view controls, seven frontend contracts and 151 regressions; 23 exact shipped-ESM DOM cases pass, including two new controls. Final runs take 4.299 s/201,863,168 and 2.689 s/314,146,816 sampled group-RSS bytes. The first accidental native-case selection remains failed; the guard blocks molecular execution and its explicit deselection is preserved. That real-provider regression remains required for actual installed receiving. No native View loading, browser, comm, canvas or independent molecular reproduction is claimed.
• The receipt and 178-member immutable archive, manifest SHA-256 9fbdb4081342dc80e117ff731837acc7be8a91edf690c157325a4e5173c4a04a, preserve all source attempts, exact source objects, native events/JUnit/TAP, guard/tool identities and terminal receipts. Isolated verified Node26/npm11.16/jsdom30.1.3 supplies the declared frontend engine; Conda/provider pins and scientific inputs remain unchanged. Runtime resources add the new adapter path; historical helper/data hashes stay intact.
• Thirteen final local checks and three fresh actual-committed-export resource/index/publication checks pass; all 207 published blobs match qualified bytes. All groups are terminal, with no resource interruption. [skip ci] preserves the maintainer route; #28 exact-head hosted evidence remains missing and no manual CI is dispatched. Private task/tool/dependency/raw custody is #50, review 2026-11-14.
• Next: identify a new exact artifact containing this route, then receive normal/pm-fs on original carriers/full results, including real provider visuals/metadata, frame/pose axes, source nonmutation and the excluded native regression. The older cfff386 wheel lacks the new route. MolSysMT #642 has owner triage but no delivered fix; receive it separately. #60, full E4/#73, other-carrier projection, E3 #604, canvas, independent science/reproduction and public/minor/platform gates remain open.

---

Earlier native coverage, installed binding and original request (preserved)

Native unevaluated-family receiving and provider feedback — 2026-11-07

Published checkpoint, registered populations and outcomes, and current handoff preserve distinct passing, failed and interrupted selections. This issue remains open/partial; the maintainer resumed the complete objective after the preceding pause.

• Native information-state receiving passes for the focused empty-source population, normal and fresh pm/fs. The unchanged original retained carrier keeps 3 H-bond/2 hydrophobic observations and both families evaluated. Public MolSysMT invalidate_structures([0]) on the translated control's already empty hbonds yields no evaluated frames; hydrophobic remains evaluated-empty. Both states survive detached accessors, queries/pages and complete H5MSM 0.5 write/readback. Original-result JSON, all original native/analysis sources and all five original poses remain exact. The separate control never becomes a sixth original pose; four original poses remain unevaluated.
• Positive-source invalidation persistence remains failed under MolSysMT #642. Two normal attempts reproduce loss of the unused hbonds.evidence_labels table at H5MSM readback; the diagnostic delta contains exactly that field. No live occurrence, coordinate or evaluated-coverage change is claimed. No downstream codec patch, stripped labels or passing exception is introduced. The empty-source pass does not clear this failed composition; pm/fs for it is unexecuted.
• The first expanded empty-source attempt is preserved as a 60.071 s wall interruption, 622,927,872 bytes sampled group RSS, with final completion/custody unreached. Its partial results do not count as passing. A separately registered focused selection completes in 36.733/37.048 s, 723,906,560/725,893,120 bytes; per profile it receives two bindings, four queries/pages, two complete carrier roundtrips and two relevant refusals. The preceding fourteen general installed refusals and 190 source cases keep their separate unchanged-byte scopes. No limit or complete-equality assertion is weakened.
• The consumer remains exact source cfff3866d511e7f1d202e3e03dd8c54a38963dc2, wheel 0.0.0+307.gcfff386, SHA-256 bc6da61abcb51a6950b2c5ed388616a70dd7cd393056c2b98d3665f9a40d1c51; borrowed MolSysMT is 9e1f1b1, Viewer 78ede79e is import-only and the original scientific producer remains 5952053. No runtime implementation, provider pin, original input, installed-test helper, shared environment or source checkout changes. No detector, preparation, molecular transformation, energy, search or View is executed.
• Primary archive, 210 members, SHA-256 b4a676022aa26f0508fadd87bb5f1dee21d9e7650edf4ebe085599c4313793ef, preserves both failures, interruption and passing profiles with frozen code/registrations, complete payloads and terminal receipts. The 17-member reader companion is 3981244aeef9c31d8a08b8d4c00100718195bf8cdf449dbc19a3ddba635bffa0. First independent reader passes in 9.112 s/325,423,104 bytes, verifies all stored/decoded members and original/derived numerical witnesses, imports no scientific package and performs no molecular operation. It preserves the declared coverage-dependent execution projection and original-source execution records; this is not producer authentication or independent molecular reproduction.
• Thirteen final local checks and three fresh actual-committed-export resource/index/publication checks pass; all 238 published blobs match the qualified bytes. All owned groups are terminal. Only the expanded attempt has a resource interruption. The commit carries [skip ci]; exact-head hosted evidence remains missing under #28 and no manual CI was dispatched. Task data stay under #50 custody, review 2026-11-14; MolSysSuite #173 retains shared guard ownership.
• Next: admit an actual single-frame addon/scene route, preserving the full result and known pose/frame axes. The existing renderer builds one frame per pose; do not silently substitute this one-frame carrier or broadcast analysis to the other four poses. Receive the #642 provider correction separately. Whole E4/#73, other-carrier correspondence, E3 #604, canvas #52/#60, independent science/reproduction and public/minor/platform delivery remain open.

---

Previous installed binding and preserved history

Installed pose binding qualified; maintainer pause — 2026-11-07

Published checkpoint records actual installed receiving and the resumption checkpoint. This issue remains open/partial; work is paused at the maintainer's request. No new scientific cell or addon work starts during the pause.

• The new private wheel 0.0.0+307.gcfff386, SHA-256 bc6da61abcb51a6950b2c5ed388616a70dd7cd393056c2b98d3665f9a40d1c51, comes from exact API source cfff3866d511e7f1d202e3e03dd8c54a38963dc2. Both normal and fresh pm/fs processes pass with real MolSysMT copies, queries and complete H5MSM 0.5 carrier write/readback. Per profile: two accepted bindings, fourteen refusals and two full-carrier roundtrips. Source qualification remains the preceding 190 cases; these new records supply actual installed/provider receiving.
• All five original result records, scores and provenance remain exactly intact. The retained original pose is public zero-based index 0; the historical wire position 1 is one-based, and the legacy atom correspondence has two columns. The separate translated evaluated-empty control retains its own derived ID/source hashes, no original score/rank and never becomes a sixth original pose. Four original poses remain unevaluated. No preparation, H generation, detector, metric, energy, search or View is executed.
• Original scientific producer 5952053, borrowed MolSysMT 9e1f1b1, Viewer import-only overlay 78ede79e and the new consumer retain their distinct identities. The 74-file new consumer and 3794-file transported provider prefix exclude 73 old consumer files; the 641-file Viewer overlay is only part of the private import composition. Shared Conda metadata, provider checkouts and pins remain unchanged. This does not qualify public dependency closure.
• normal-03 passes in 29.635 s/638,640,128 bytes sampled group RSS, pm-fs-01 in 29.703 s/637,353,984 bytes. CPU0/nice10/one worker and all registered finite bounds hold; no resource interruption or live owned descendant remains. These instrumented observations are not API throughput claims.
• The 119-member primary archive has manifest SHA-256 0dc17c3dd7c6e2db8f03ca0756603a016694537195394dc758c6909b34592c45; the 34-member reader companion is ee54ee5e6d42e5e1e0b25584cc40544f07d57e79d80ae4a3a436fd7c2706685f. Two initial client parameter failures and two overstrict reader failures are preserved. Each complete per-profile analysis payload exactly matches its older immutable witness. Thirteen preexisting floating measurement differences and distinct control-source identities are preserved; cross-profile numerical agreement uses explicit length/angle units and absolute tolerances, not rounded replacement records.
• Independent standard-library reader-03 verifies all 119 stored/decoded members and the complete original results in 18.526 s/272,764,928 bytes, with no scientific import or molecular operation. This is record receiving, not independent molecular reproduction or producer authentication. Thirteen final local checks and three fresh actual-committed-export resource/index/publication checks pass; all 164 published blobs match the qualified bytes. The commit carries [skip ci]; exact-head hosted evidence remains missing under #28, and no manual CI was dispatched.
• Next after resumption: register and receive a native named interaction family with zero evaluated frames through supported MolSysMT operations; source False-coverage doubles do not establish this native case. Then admit the actual single-frame addon/scene binding with known axes, preserving the complete result and refusing broadcasts to other poses/carriers. Whole E4, other-carrier correspondence, deferred canvas #52/#60, E3/MolSysMT #604, independent science/reproduction, public delivery and full #73 remain unaccepted. Private candidate/worktree/prefixes remain under #50 custody, review 2026-11-14; MolSysSuite #173 owns shared execution support.

---

Previous API implementation and original request (preserved)

Public pose–carrier binding implemented — 2026-11-07

Source checkpoint publishes bind_interactions/PoseInteractions, its public contract and limits, and source qualification. The issue remains open/partial. The earlier saved-carrier receiving and this new API's source tests retain separate producer/scope identities.

• An explicit zero-based pose position, ordered atom map, known state IDs and common-frame declaration associate one selected pose with its admitted one-structure carrier. Ordered mapped labels/elements and unit-bearing coordinates are checked, including nonmonotonic maps; position/rank/pose ID remain distinct. Default absolute tolerance is 1e-12 angstrom, relative tolerance zero; overrides require an explicit finite nonnegative scalar length.
• The whole original result is preserved as a detached snapshot. Result, selected-pose, analysis and carrier accessors do not rewrite the source or retained binding. Provider copy/interchange and evaluated-frame coverage remain MolSysMT-owned; empty and unevaluated analyses are not conflated. No other original pose becomes evaluated. No molecular matcher, preparation, detector, alignment, scoring or search is introduced.
• State/frame labels are caller declarations, not chemical equivalence, unmapped receptor-geometry validation or producer authentication. The binding accepts an already admitted carrier. Missing maps/states/analyses, wrong identities/coordinates, invalid axes and multi-frame carriers are refused; no analysis is projected/broadcast to another scene. Its association dictionary is inspection evidence, not a molecular persistence codec.
• 190 complete source cases pass: 44 new protocol-double controls and 146 existing argument/model/coordinate/result regressions. Five final controls additionally cover nonfinite mapped coordinates, provider-error propagation and extra original key annotations. The worker forbids real MolSysMT operations and Vina import. Both initial fixture failures, the earlier 185-case pass and final complete selection are retained. Actual native-system/installed-artifact compatibility is not established by these doubles.
• The final selection completes under CPU0/nice10/one worker in 2.762 s with 128,102,400 bytes sampled group RSS, no reserve interruption and no live owned descendant. These are source-test resource observations, not API throughput measurements. Thirteen final local checks and three fresh actual-committed-export resource/index/publication checks pass; all fifteen published blobs match checked bytes. The commit carries [skip ci]; exact-head hosted evidence remains missing under #28 and no manual CI was dispatched.
• The initial collection error also reproduces known Pytest Receptor #46. The compact header miscounts executed cases while exit 2/COLLECTION_ERROR are correct. Native pytest events, JUnit, process status and completion receipts govern this qualification; the existing provider issue receives the consumer evidence.
• Next: register an actual isolated candidate/source for the new API and receive the original carriers normal first, pm/fs second, including real provider copies/persistence and positive/empty/unevaluated/refusal/nonmutation controls. The old a7422df private consumer lacks this API. Addon/scene axes, deferred canvas, other-carrier correspondence, public delivery, independent science and full #73 keep their own gates.

---

Previous saved-carrier receiving and original request (preserved)

Published saved-carrier receiving — 2026-11-07

Source checkpoint, receiving report and frozen passing notebook are now published. This issue remains open/partial: the same-carrier consumer composition passes; no new DockingMT binding API, addon/canvas or whole-E4 acceptance is claimed.

• Both normal and fresh pm/fs processes receive the two original complete 3238-atom/one-frame carriers through public MolSysMT and real Python Viewer APIs. Each closes two views, preserves full named analysis payloads, receives filtering/pages, independent attachment/loading and interactions-only H5MSM persistence/readback, and passes eleven refusals. Queries exercise MolSysMT Interactions.between_selections; newly saved layers are reread with msm.h5msm.read_layers, not loaded into another new View.
• The original five-pose result and scientific producer remain intact. Only the retained geometry admits the original thirteen pose atoms at carrier indices 3220–3232. The translated control remains distinct and four other original poses stay unevaluated. Six named aliases do not become six scientific observations.
• Actual bounded profiles pass in 31.344 s/552.52 MiB and 31.562 s/554.73 MiB sampled owned-group RSS, with 50 typed exports/profile, all finite caps, original source/payload nonmutation and zero new preparation/H/charge/type assignments, detectors, metrics, energies or searches. These are instrumented receiving measurements, not portable performance claims. No resource interruption or live owned descendant remains.
• The 238-member archive, manifest SHA-256 73c8114368f0d7011d2e27d229662bf29be1f99bea27c0e0a24065f077e04b4d, retains the first diagnostic-client failure, corrected passing profiles, guard controls and administrative sealing error. The nine-member reader companion, a3cf0e74acd65821a06aa57f859cf009d71f7851cd33f3fe95ee4b698f7fa3f3, passes detached standard-library record receiving in 2.856 s/206.85 MiB. It verifies all stored/decoded records and numeric witnesses without molecular imports/restoration or new calculations; it does not authenticate the producer.
• Passing notebook SHA-256 is a6a8561e4aa7d9d2b5967fd43b9c2fdf8793ba13bbbb3c3a5d524e3d53e7d113. Consumer stays a7422df/wheel 93d76f89, borrowed MolSysMT 9e1f1b1, Viewer 78ede79e/wheel 3d31a4d9 and original scientific producer 5952053. Qualification base f042244 and publishing source retain separate identities. No installation, provider pin or sibling-worktree change occurs.
• Thirteen applicable local checks and three fresh actual-committed-export resource/index/publication checks pass, with all 258 published blobs matching the checked bytes. The commit carries [skip ci]; required exact-head hosted evidence remains missing under #28. No manual workflow was dispatched. The maintainer resumed the full objective after the preserved pause; the original heavy interruption and deferred canvas review remain separate.
• Next implement the smallest docking-specific association between an explicitly selected original pose and its admitted carrier, preserving the full result and refusing unknown/incompatible scene mappings. MolSysSuite #173 owns reusable execution-guard support. E3 #604, other-carrier correspondence, usable addon/canvas, independent scientific panel, public delivery and full #73 keep their own gates.

---

Previous source-review checkpoint and original request (preserved)

Saved-analysis consumer admission checkpoint — 2026-11-07

The source-reviewed public recipe and maintained owning report are published at f04224449fe5d9a5bb7b8e61a1b4e5435ceb2103. This is preparation for the remaining consumer/addon receiving, with no new API or molecular execution.

• Correct the carrier arithmetic in the maintained handoff: original normal/pm-fs associations retain ligand indices 3220..3237, hence analysed receptor_h3220 + ligand18 = 3238 atoms. The prepared/PDBQT receptor2011 belongs to another population. No current destination scene size/order is inferred; thirteen original ligand atoms and five added nonpolar H retain their original witness/provenance.
• Public complete-carrier/layer reads, carrier extraction, full-result remap and Viewer attach/load exist at the original qualified sources MolSysMT 00abb958f and Viewer 78ede79e. Source signatures do not qualify this new composition. assume_aligned=True declares matching axes/geometry/boxes; it does not authenticate them, and the loader does not embed a subsystem into a larger scene.
• First receive the same complete saved carrier as a baseline. A different docking scene needs admitted receptor/ligand/frame correspondence and physically checked retained coordinates before attachment. Preserve complete persistence, evaluated/unevaluated coverage, participant loss and empty/unsupported distinctions. Repeating the one saved frame cannot label the other four poses as evaluated.
• Not ready for execution: fix destination/maps, supported calls and loss/coverage controls, then seal exact scripts/artifacts, finite resources/call caps and venue under the existing local hold. No H generation, detectors, scoring, searches, local molecular replay or manual hosted CI is started by this increment.
• All thirty preserved original E4 exports and all 193 E2 archive members pass byte/hash checks during closeout. Thirteen applicable local checks and three fresh actual-committed-export checks pass; skipped hosted evidence remains missing under #28. The previous molecular receiving keeps its bounded passing scope, while whole E4, usable addon/canvas, public delivery and #73 remain unproven.

---

Earlier receiving and original request (preserved)

Concrete E4 workflow

#73/DMT-046 requires basic structured interaction analysis within complete Core+Extended functional delivery before stabilization. This issue owns receiving and composing that analysis for existing docking poses, using MolSysMT as the molecular-science provider.

For a selected pose of one known ligand/receptor state, use a provider-supported assembled molecular system in the original receptor Cartesian frame. Evaluate explicit disjoint ligand and receptor selections with pbc=False, a selected structure and named MolSysMT criteria. The first profile considers hydrogen bonds (baker_hubbard/nitrogen_oxygen) and hydrophobic typed proximity (atom_pair_distance/smarts_hydrophobic_atoms), preserving each family's exact inequalities, units, assumptions and experimental scope. This does not require all interaction families or establish interaction energies/binding validity.

Existing provider boundary

Static inspection of MolSysMT fetched main ed09f64539a5af6c1dbe9bd352c7ef65dcd4a7a6 finds public interactions.hbonds.get_hbonds, interactions.hydrophobic.get_hydrophobic_interactions, typed molsysmt.Interactions, evaluated coverage, participant/structure source maps, independent query/persistence and original method/producer attribution. #252 is delivered. Reuse these; no missing-detector claim or duplicate proposal is warranted from this inspection. This source is separate from qualified DockingMT runtime pin 00abb958f and from public artifact qualification.

PharmacophoreMT owns pharmacophore feature/model interpretation, not generic replacement of the existing MolSysMT detectors. MolSysViewer owns rendering/communication, not interaction definitions or the scientific analysis.

Work and acceptance

1. Receive existing public provider operations on the actual qualified dependency/runtime. Prove ligand/receptor assembly, pose/source atom maps, chemical state, hydrogen inventory and structure association using supported molecular operations. Report any missing reusable route in MolSysMT; no local merger, coordinate rewriter, chemical typing, detector or distance engine.
2. Preserve the provider's structured analysis, original method/profile/cutoffs, evaluated-empty versus unevaluated/unsupported states, units, participant roles/source indices, state/structure coverage, assumptions, bibliography and producer versions. Do not infer “no biological interaction” from absent H, unsupported chemistry or an empty evaluated model.
3. Associate analysis with the actual original DockingResult/DockingPose, receptor/partner state and coordinate/input identities. Never overwrite original scores or relabel a saved producer. Choose a supported provider interchange/persistence route; do not clone its molecular codec or schema in DockingMT.
4. Demonstrate provider-backed filtering/query and a finite representative pose workflow with explicit positive/adverse/empty controls, nondefault units, source nonmutation, exact producer/consumer provenance, bounded CPU/RAM and detached receiving. Register computation/population/budgets before execution. Molecular numerical correctness and consumer association both need evidence.
5. Keep the structured information usable by the addon through supported provider contracts; actual rendering/live-canvas acceptance remains #52/#60 and the maintainer's scheduling decision. A Python object, static export or documentation is not visible acceptance.

Only docking-specific population selection, association, filtering policy and interpretation belong here. Existing molecular bridges retain #49 ownership. Public/platform, independent scientific-panel and full-release gates remain separately required. Source inspection/planning does not accept E4 or authorize a provider pin change, new native docking campaign or automatic heavy restart.