feat(abi): MiniMax-H3 video+audio generation through the ONE surface — vllm_video_*, ABI v12, /v1/videos via the library seam, thin-client examples - #123
Closed
localai-bot wants to merge 1 commit into
Conversation
mudler
force-pushed
the
row/H3-VIDEO-ABI
branch
from
August 7, 2026 22:00
900662c to
3e055cf
Compare
…— vllm_video_*, ABI v12, /v1/videos via the library seam, thin-client examples ONE-SURFACE fold ROW 2 (row/H3-VIDEO-ABI, task #283; the #121 precedent's shape: three-arm fold gate, guard-ratchet shrink). The library absorbs everything the examples privately owned. GOLDENS FIRST: minimax_h3_video_fold_fixture.h writes a deterministic tiny checkpoint set (857KB F32 ComfyUI-GGUF DiT at the test_minimax_h3.cpp:3786 reduced geometry, on-disk reduced ViT3D/BigVGAN VAEs + config.json, prompt embeds); the PRE-fold minimax-h3-gen binary AT THE BRANCH BASE (fc636c7) rendered it (--partition fl2va --steps 3 --frames 5 --height/width 32, CPU, keep-quant; run-to-run deterministic, diff -r clean) and its 8 frames + WAV + both minimax-h3-mux --print-only argv lines are the committed goldens in fixtures/minimax_h3_video_fold/. W1 seam: vllm::multimodal::MiniMaxH3VideoEngine absorbs the 1293-line example driver AND the server's 354-line /v1/videos twin — 4 DiT loader arms (GGUF keep-quant / dequant-bf16 host+streamed / bf16-shard stream / NVFP4 bf16+fp4-resident), VAE decoder + lazy encoder halves, the H3-Encoder tower staged once (GGUF or bf16 shards), the #77 partition guard, fl2va/ref2va conditioning, the byte-exact splitmix64 Box-Muller noise streams (VT_H3_GAUSSIAN_NOISE honored; seeded requests derive the audio stream via splitmix64), artifact writing and mux-argv assembly. The library SPAWNS NOTHING (mkdir -p became std::filesystem; ffmpeg stays caller-side — the ratified 2026-08-03 boundary). MiniMaxH3VideoGenParamsFromRequest is the ONE /v1/videos mapping so HTTP and FFI cannot drift. W2 ABI: the ratified vllm_video_* slice (one-surface-abi.md:92-108) — vllm_video_engine_load/free, vllm_video_model_params/_default, vllm_video_params/_default, vllm_video_generate, vllm_video_result/_free — plus the engine-free vllm_video_mux_argv(+_params_default,+_free) composer the mux thin client needs. VLLM_ABI_VERSION 11 -> 12; test_capi floor pin >= 12 (the == checks move with the macro). As-shipped deltas vs the proposal argued in the spec (+prompt_embeds_path/+partition/+fp4_resident/+output_dir; -task/-duration; ONE ref_image). Zero values preserve behaviour. Refusals both directions: vllm_video_engine_load on a text dir names vllm_engine_load; vllm_engine_load on the H3 dir fails byte-for-byte as captured at v11. W3 server: the VideoState block is DELETED; /v1/videos = seam Load + a runner lambda (FromRequest -> Generate -> fork/execvp(mux_argv)). Direct MiniMaxH3 refs 54 -> 7, all seam type names. DISCLOSED numeric deltas on the server arm (no goldens existed; the drift WAS the defect): legacy single-stream uniform noise -> the shared ratified recipe; host-f32 GGUF default -> keep-quant; new --video-dequant-bf16 flag. W4 examples: minimax_h3_gen (1293 -> 216 lines) + minimax_h3_mux are vllm.h+vllm::shared thin clients, BOTH byte-identical to the pre-fold binaries on the fixture (frames+WAV cmp clean; mux argv diff clean; the pre-fold diagnostic modes died with the private pipeline — named residual). Ratchet: both example-abi-allowlist rows removed, MAX_INTERNAL_REACHING 11 -> 9 with the equality pin + spec claims moved; the abi-capability video row closes; FEATURES video row -> reachable naming the four symbols. Gates: three-arm fold gate test_minimax_h3_video_fold 4 cases / 131 assertions GREEN (seam == replicated pre-fold pipeline == committed goldens, byte-identical; partition-guard + input refusals; FromRequest field-complete); test_capi 40/373 incl. the v12 golden e2e THROUGH the C marshalling; vllm_capi_c_check strict-C11 green; test_minimax_h3 75/75 unchanged; test_openai_api_server 45/566 incl. the socket-level routes-do-not-exist-without-a-runner 404 pin; test_video_api 14/14; check-surface-coverage green with the SHRINK enforced + its 46-test mutation suite. GB10 real-video re-verification through the v12 ABI is a NAMED RESIDUAL (box on the Kimi campaign); records updated (STATUS/BENCHMARKS/FEATURES/NOW/state/model-matrix + both specs). FOLLOWING_AGENTS_PROTOCOL Assisted-by: Claude Code:claude-fable-5 [ClaudeCode]
mudler
force-pushed
the
row/H3-VIDEO-ABI
branch
from
August 7, 2026 22:05
3e055cf to
3d501c6
Compare
mudler
added a commit
that referenced
this pull request
Aug 7, 2026
…- vllm_video_*, ABI v12, library-owned /v1/videos, thin-client examples (#123) ONE-SURFACE fold ROW 2, the largest: video generation becomes reachable to embedders (LocalAI's vllm-cpp backend is the motivating consumer). The library absorbs the 1293-line example driver into a video seam; include/vllm.h gains the ratified slice - vllm_video_engine, vllm_video_model_params/_default, vllm_video_params/_default, vllm_video_generate, vllm_video_result (frame dir + WAV + mux argv: the library COMPOSES the ffmpeg command, the caller execs, src/ spawns nothing), VLLM_ABI_VERSION 11->12 floor-pinned. The server's /v1/videos routes through the same seam (VideoRunner library-owned; ~54 direct MiniMaxH3 refs dropped); minimax_h3_gen + minimax_h3_mux are rewritten as vllm.h-only thin clients; surface-guard ratchet 11->9. Correctness: three-arm fold gate on a committed deterministic fixture - library seam == replicated pre-fold pipeline == goldens captured from the pre-fold binary at the branch base - frames+WAV byte-identical through the seam, the C marshalling, and the rewritten binaries. DISCLOSED numeric change on the server arm only: pre-fold /v1/videos had drifted to single-stream UNIFORM noise + f32-host default while the ratified example used the Gaussian (torch.randn-mirror) recipe - review-confirmed by archaeology at the base SHA; the fold unifies the server onto the ratified behavior (the drift WAS the defect the ONE SURFACE directive exists to prevent). ABI-shape deltas vs the spec proposal argued in the spec's "As-shipped deltas". Two-round-capable review: fresh mutation review (seam and replica arms independently killable, ABI floor pin, zero-value contract, route gating, ratchet pin, thin-client purity, scope) returned MERGE-CLEAN with 3 minor doc findings, recorded as follow-ups. Operator re-ran fold/h3/capi/video_api/api_server on this merged tree: 5/5 SUCCESS. Residual: GB10 real-video re-verification through the v12 ABI (box running the Kimi campaign). FOLLOWING_AGENTS_PROTOCOL Assisted-by: Claude Code:claude-fable-5 [ClaudeCode]
Collaborator
Author
|
Landed on main as c0bab82 (mudler-authored squash via commit-tree). Review: fresh mutation review MERGE-CLEAN (seam and replica arms independently killable; the server numeric-delta claim CONFIRMED by archaeology at fc636c7 — pre-fold /v1/videos used single-stream uniform noise + f32-host while the ratified example used the Gaussian recipe; the fold unifies onto ratified). Operator-run gates on the merged tree: keyed auto-merges verified byte-identical to the deterministic rebuild; the state-entry extraction was rebuilt after a first-pass slice error (caught by seam inspection, not the checker); fold/h3/capi/video_api/api_server 5/5 SUCCESS; eleven gates 00000000000 after a 40-char STATUS compaction within this PR's own row (merge composition tipped the shrink-only ratchet). Video generation is now on the one surface: ABI v12, ratchet 9, capability table 9 reachable / 2 tracked. 3 minor review findings (stale c_header_compile claim, v12 changelog placement, fixture-flag doc note) recorded as follow-ups for the next row. Residual: GB10 real-video re-verification once the box frees. |
mudler
added a commit
that referenced
this pull request
Aug 7, 2026
…nner; engine==CLI 128/128, golden 122/128 profile; ABI v13 vllm_complete_tokens; thin-client example (merged ratchet 8) Branch: row/KIMI-RUNNER-FOLD (draft PR #122 — the helper claim). ARCH-ONE-SURFACE ROW 7 / kimi-linear.md §20.3→§21, task #281. The runner previously ABORTED on Kimi's KV (MambaSpec check vs config-derived zeros); the fast paged-incremental decode was CLI-only. Now Kimi decode runs THROUGH ModelRegistry::Forward on the runner's own paged state and the engine/server serve it. B1 — KV enablement: LoadHfConfig synthesizes layer_types + the GDN-group geometry from Kimi's nested linear_attn_config (configs/kimi_linear.py:34-148, 1-indexed kda_layers). ADDITIVE: explicit-field configs (qwen3_5) never enter the branch — the priority conjunct is PINNED by a both-present test (explicit fields WIN; mutation-verified RED when the conjunct is dropped); runner.cpp untouched. B2 — KDA-paged block (KdaLayerPagedBf16): vt::KdaChunkPrefill for fresh prefills (VT_KIMI_PAGED_KDA_CHUNK=0 A/B) / vt::KdaGatedDeltaRule (T==1) for decode + continuing prefills over the paged gdn_state group keyed by non_spec_state_indices (GdnStateGather/Scatter); conv taps via CausalConv1dFwd (varlen) / CausalConv1dUpdate (decode) in vLLM's conv_state.chunk(3) [q|k|v] layout. NOT per-head GdnBlockPaged — KDA's per-K-channel decay needs the KDA ops; shared GDN kernels untouched. B3 — NoPE-MLA-paged block: latent rows via vt::ConcatAndCacheMla at attn_meta.slot_mapping (bf16 pages; conv cache dtype follows ResolveKvCacheDType, mirroring kda_state_dtype). PRODUCTION arm = mla::ForwardMlaAttentionBlock (vLLM's absorbed-MQA decode / FA2 prefill, identity-RoPE, scale qk^-0.5, load-time AbsorbKvBProjBf16 into MlaResidentWeights::w_uk_t/w_uv) — VT_KIMI_PAGED_MLA_FA2 DEFAULT ON, GB10-ruled: it reproduces the golden's 122/128 near-tie profile EXACTLY. Diagnostic arm ('0') = the exact f64 island (CPU fold-identity vehicle; GB10 111/128 = the §19 GPU M-tiling near-tie class, not a paging bug). B4 — ONE SURFACE: registry loads the bf16-RESIDENT tower through the engine (ModelFactory::stage_on_load — queue before load, per-tensor stage-and-release, the §13 recipe); ForwardPaged honors device_token_ids (the DEFAULT-ON async device mirror leaves host ids stale — a measured GB10 9/128 divergence, RED-first CPU-pinned); vllm_complete_tokens (ABI v13 after the #123 video-v12 rebase; changelog slotted; floor pin >= 13; dlopen 28/28 + c_header_compile reference); examples/kimi_linear_gen REWRITTEN as a thin vllm.h + vllm::shared client; example-abi-allowlist kimi row REMOVED — merged ratchet MAX_INTERNAL_REACHING = 8 (with #123's two minimax removals), equality pin 8, coupled spec/state claims moved together; the CLI-incremental reference leg preserved as the env-gated test_kimi_linear_fold_gate. GB10 gates (dgx, /dev/shm CUDA build: CUTLASS 4.5.0 + FA2 + Triton AOT sm_121a; golden md5 bfa5bdbf; flock, drop_caches, worker parked, no reboot during the gates): - SACRED post-fold: 35B test_qwen36_paged_engine 2/2·315 PASS; 27B test_qwen27_paged_engine 1/1·235 PASS. - Gate A fold identity: CLI reference leg reproduces §19 EXACTLY (122/128 @ 18.93 tok/s, p7 got byte-equal); ENGINE leg (vllm_engine_load + vllm_complete_tokens, FA2 default) == CLI 128/128 BYTE-IDENTICAL; vs golden 122/128 — the SAME near-tie profile (>=122 bound MET). - Server smoke /v1/completions: PASS — streamed 48 tokens / 2.52 s = 19.0 tok/s wall, coherent; non-streamed coherent; tokenizer via TikTokenConverter (round-trip verified; Kimi ships tiktoken-only). - SPEED: server 19.0 tok/s wall (~0.90x the #111 vLLM ~21 floor) — >= vLLM NOT met; residual: per-step KDA host islands, grouped MoE, decode graph (spec §21). Example two-length diffs (16.9 N=64) carry cold-start pollution — measurement caveat recorded. - The in-session vLLM re-measure ABORTED BY BOX REBOOT (util 0.82 at graph capture — the §19-documented risk; not retried per the safety mandate; the #111 ~21 floor stays the denominator; box recovered clean, worker auto-restored). - CPU: test_kimi_linear_paged 8/8·206 (ForwardPaged logits BYTE-EQUAL the CLI at every step, mutation-RED on 4 seams; slot isolation; batched-prefill slot writes; device_token_ids pin; FA2-arm greedy == exact-arm + f32-page rejection; pid-unique temp fixtures); test_hf_config 18/18·187; test_capi 400; test_dlopen 28/28; full ctest 351/351 (pre-rebase base). Records: spec §21, STATUS/BENCHMARKS/FEATURES keyed rows, model-matrix row, state.md append + NOW refresh, benchmark-record append, ENVIRONMENT.md knobs (VT_KIMI_PAGED_MLA_FA2 documented default-ON, GB10-ruled). FOLLOWING_AGENTS_PROTOCOL Assisted-by: Claude Code:claude-fable-5 [ClaudeCode]
mudler
added a commit
that referenced
this pull request
Aug 7, 2026
…odel_params.device, ABI v14, --device on server + cli ONE-SURFACE fold ROW 8 (row/DEVICE-KNOB, task #284), the smallest: an embedder can now ASK for a device instead of inheriting the accelerator-first probe. Mirrors vLLM's DeviceConfig.device names (vllm/config/device.py:13 @ 555967922; an explicit device is assigned VERBATIM and never substituted, device.py:61-66 - upstream's own --device flag was removed in v0.10, bc8a8ce5e, so the field mirrors the surviving config surface). include/vllm.h gains vllm_model_params.device (int32, 0=auto/1=cpu/ 2=cuda; 0 MUST be auto so zero-initialized structs keep the pre-v14 probe byte-identical; cpu-before-cuda follows the shipped v12 vllm_video_model_params.device precedent shifted by the auto slot). VLLM_ABI_VERSION -> 14 per the operator's version-collision heads-up (#122 reconciles the Kimi vllm_complete_tokens slice to v13 at its landing; the changelog names the gap and this branch rebases after). Floor pin advanced to >= 14 (the #121 == lesson). New vllm/config/device.h (+.cpp): vllm::Device, DeviceFromString, DeviceName. Plumb: EngineParams::device -> SelectQueue explicit arms via the pure LoadedEngine::ResolveExplicitDeviceType (explicit cpu NEVER probes; explicit cuda requires the registered kCUDA platform and fails LOUD otherwise; explicit-arm queue-creation failures propagate, unlike auto's byte-identical catch-and-fall-back). FromModelDir resolves an explicit device BEFORE any path/config I/O (DeviceConfig resolves at config time, arg_utils.py:1878), so a device error is never masked by a path error. capi validates 0/1/2 and refuses explicit cuda on the CPU-hosted transcription stack. Thin clients: --device auto|cpu|cuda on vllm-cli (ABI field only) and the server (DeviceFromString; --video-device untouched, separate engine). Also folds the #123 review's three MINOR findings (tests/docs only): c_header_compile.c actually references the v11+v12 surface + the v14 field (its "every entry point" claim went stale at v10); the v12 changelog block moved to chronological position; the H3 fold fixture flag list gained --keep-quant. Gates: full CPU build -Werror clean; test_capi (4 new v14 cases + floor), test_loaded_engine_dense (+3 device cases incl. the pure policy matrix that pins "explicit cpu beats a REGISTERED accelerator" CPU-side), test_dlopen, test_openai_api_server (+explicit-cpu serving smoke), test_minimax_h3_video_fold, test_model_loader_gguf, test_platform all green; surface-coverage green, ratchet UNTOUCHED at 9; STATUS ratchet paid by removing a stale back-to-back merge duplicate (279200 -> 279150). Ten mutations killed and reverted (defaults, zero-map, range, both plumb sites, ctor arm, unconditional fallback, policy flip, parse flip, forgotten bump); kills recorded in .agents/state.md. Residuals: CUDA-build A/B (no dgx in scope) and the pending #122 rebase, both named in the state entry. FOLLOWING_AGENTS_PROTOCOL Assisted-by: Claude Code:claude-fable-5 [ClaudeCode]
mudler
added a commit
that referenced
this pull request
Aug 7, 2026
…odel_params.device, ABI v14, --device on server + cli ONE-SURFACE fold ROW 8 (row/DEVICE-KNOB, task #284), the smallest: an embedder can now ASK for a device instead of inheriting the accelerator-first probe. Mirrors vLLM's DeviceConfig.device names (vllm/config/device.py:13 @ 555967922; an explicit device is assigned VERBATIM and never substituted, device.py:61-66 - upstream's own --device flag was removed in v0.10, bc8a8ce5e, so the field mirrors the surviving config surface). include/vllm.h gains vllm_model_params.device (int32, 0=auto/1=cpu/ 2=cuda; 0 MUST be auto so zero-initialized structs keep the pre-v14 probe byte-identical; cpu-before-cuda follows the shipped v12 vllm_video_model_params.device precedent shifted by the auto slot). VLLM_ABI_VERSION 13 -> 14 on top of #122's vllm_complete_tokens v13 (rebased onto 9f772ca; the changelog now reads v10..v14 chronological - the v12-after-v11 move is the #123-review minor kept through the rebase). Floor pin advanced to >= 14 (the #121 == lesson). New vllm/config/device.h (+.cpp): vllm::Device, DeviceFromString, DeviceName. Plumb: EngineParams::device -> SelectQueue explicit arms via the pure LoadedEngine::ResolveExplicitDeviceType (explicit cpu NEVER probes; explicit cuda requires the registered kCUDA platform and fails LOUD otherwise; explicit-arm queue-creation failures propagate, unlike auto's byte-identical catch-and-fall-back). FromModelDir resolves an explicit device BEFORE any path/config I/O (DeviceConfig resolves at config time, arg_utils.py:1878), so a device error is never masked by a path error. capi validates 0/1/2 and refuses explicit cuda on the CPU-hosted transcription stack. Thin clients: --device auto|cpu|cuda on vllm-cli (ABI field only) and the server (DeviceFromString; --video-device untouched, separate engine). Also folds the #123 review's three MINOR findings (tests/docs only): c_header_compile.c actually references the v11+v12 surface + the v14 field (composed with #122's vllm_complete_tokens reference; its "every entry point" claim went stale at v10); the v12 changelog block moved to chronological position; the H3 fold fixture flag list gained --keep-quant. Gates: full CPU build -Werror clean; test_capi (4 new v14 cases + floor >= 14, #122's v13 cases green alongside), test_loaded_engine_dense (+3 device cases incl. the pure policy matrix that pins "explicit cpu beats a REGISTERED accelerator" CPU-side), test_dlopen, test_openai_api_server (+explicit-cpu serving smoke), test_minimax_h3_video_fold, test_model_loader_gguf, test_platform all green; surface-coverage green, allowlist ratchet UNTOUCHED at 8 (post #122); STATUS ratchet paid by removing a stale back-to-back merge duplicate (279200 -> 279150, measured 279111 post-rebase). Ten mutations killed and reverted (defaults, zero-map, range, both plumb sites, ctor arm, unconditional fallback, policy flip, parse flip, forgotten bump); kills recorded in .agents/state.md. Residual: CUDA-build A/B (no dgx in scope), named in the state entry. FOLLOWING_AGENTS_PROTOCOL Assisted-by: Claude Code:claude-fable-5 [ClaudeCode]
mudler
added a commit
that referenced
this pull request
Aug 8, 2026
…params.device, ABI v14, --device on cli+server (#136) ONE-SURFACE fold ROW 8. vllm_model_params gains an appended device field (0=auto = today's accelerator-first probe, byte-preserving for zero-initialized callers; 1=cpu, 2=cuda), mirrored from vLLM DeviceConfig (config/device.py @ 555967922, ported as include/vllm/config/device.h + device.cpp; upstream removed the --device FLAG in v0.10 while keeping the field surface - disclosed with citation). Plumb: vllm_model_params -> EngineParams::device -> SelectQueue via pure LoadedEngine::ResolveExplicitDeviceType - explicit cpu never probes, explicit cuda on a build without CUDA fails loud with NO fallback, resolution happens before any I/O. --device auto|cpu|cuda on vllm-cli and examples/server (ABI-field-only thin clients; --video-device untouched). VLLM_ABI_VERSION 13->14, floor-pinned >= 14. Also closes the three #123 review minors: c_header_compile.c references the v11+v12+v13+v14 surface, the v12 changelog block moved chronological, the fold fixture flag list gained --keep-quant. Review: implementer 10-kill mutation battery, then a FRESH mutation review MERGE-CLEAN - zero-contract/no-fallback/never-probes/ resolve-before-IO/floor-pin all independently re-killed, the device.py port verified verbatim against the parity pin, blast radius (model_loader shared with the Kimi paged load) green. Disclosed honestly: the 0-vs-explicit-cpu distinction is CPU-tier-invisible by construction - the CUDA-build A/B is a named residual. Operator: merged tree byte-identical to the reviewed head, capi/dlopen/loaded_engine/api_server 4/4 SUCCESS. FOLLOWING_AGENTS_PROTOCOL Assisted-by: Claude Code:claude-fable-5 [ClaudeCode]
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
ONE-SURFACE fold ROW 2 (task #283) — IMPLEMENTED; single squashed commit per the doc-checkpoint per-commit obligation. DRAFT until the operator's review rounds.
What landed
minimax_h3_video_fold_fixture.hwrites a deterministic tiny checkpoint set; the PRE-foldminimax-h3-genat the branch base (fc636c7) rendered it and its 8 frames + WAV + bothminimax-h3-mux --print-onlyargv lines are committed goldens (run-to-run deterministic,diff -rclean).vllm::multimodal::MiniMaxH3VideoEngine(+MiniMaxH3VideoGenParamsFromRequest): absorbs the 1293-line example driver AND the server's 354-line /v1/videos twin. Library spawns nothing (ffmpeg boundary stands; mkdir via std::filesystem).vllm_video_engine_load/free,vllm_video_model_params(_default),vllm_video_params(_default),vllm_video_generate,vllm_video_result(_free)+vllm_video_mux_argv((_params_default)/_free); test_capi floor >= 12; refuse-both-directions pinned (video-load on a text dir names vllm_engine_load; vllm_engine_load on the H3 dir fails byte-for-byte as at v11). As-shipped deltas vs the ratified proposal argued in.agents/specs/one-surface-abi.md.vllm.h+vllm::sharedthin clients, byte-identical to the pre-fold binaries on the fixture. Allowlist −2 rows,MAX_INTERNAL_REACHING11→9 + equality pin, abi-capability video row closed, FEATURES video row → reachable.Gates (all green on 3e055cf)
test_minimax_h3_video_fold: 4 cases / 131 assertions — seam == replicated pre-fold pipeline == committed goldens, frames+WAV byte-identical; mux argv pinned.test_minimax_h375/75 unchanged.test_capi40 cases / 373 assertions incl. the v12 golden e2e THROUGH the C marshalling + zero-value contract + refusals;vllm_capi_c_checkstrict-C11 green.test_openai_api_server45/566 incl. the socket-level text-server-404s-/v1/videos pin;test_video_api14/14.scripts/agent-preflight.shEXIT=0;check-surface-coverage+ 46-test mutation suite green with the shrink enforced.🤖 Generated with Claude Code
https://claude.ai/code/session_014fZAcg1WcU8V629k6HWKys