Every same-step attempt resolves its bound route and reconciles the accepted rendered prompt against the current surface before deriving history. Preserve assembly, pre-step admission and entered users exactly once; compaction retries must not resurrect an older surviving prompt.
Normalize a broken request series to the current head plus dormant empty later system nodes, even if later versions survive or the effective text is unchanged. This removes the later-survivor exception without a delete operation and avoids appending instructions after already-admitted users on retry. Capture surface generation after reconciliation so subsequent ordinary retries do not emit phantom series headers.
Tests: both compaction cases fail before the fix (stale head, or stale later survivor), then pass with real loop/MockAdapter, pi-ai conversion and source-event reconstruction. Assert two retries keep one assembly/pre-step/user admission, preserve rendered text across section changes, and log one series boundary. Add unchanged-text explicit/tool series regressions. 391 loop tests with exact agent.ts/runtime-context.ts coverage pass at 100%; final admission spec has 11 passing cases. Source tsc and focused lint pass. Update EN/ZH implementation and owning rationale plus lifecycle retry documentation.
Resolve agent/request and prepareCall inside the accepted open step before committing system and user input. Preserve prompt-assembly model selection, bind capability and dispatch to the same prepared adapter, then log the envelope and derive the request synchronously. Preparation cancellation leaves a balanced empty step with no admitted input.
Normalize incapable-route retained prompt versions with logged per-node empty replacements and a current head, including unchanged rendering and resumed history. Dormant empty tails do not override the effective prompt. No new event or delete operation.
Tests: 389 agent-loop cases pass with exact agent.ts/runtime-context.ts coverage at 100% for all metrics; tsc -b packages/core/agent-loop/tsconfig.json passes. Seven actual-loop MockAdapter cases exercise pi-ai conversion, source-event reconstruction, both route transitions, resume, cancellation barriers, and bound selection. Stale-capability negative control fails all three transition regressions. Update EN/ZH request visibility, lifecycle generator, and owning rationale. Broad catalogs, recorded SDK/GUI and aggregate gates remain parent-owned.
Consolidate the existing in-history capability, loop, UI and artifacts into one baseline. Preserve the reviewed tree so each independently verified correction has a subsequent rationale-rich commit.
Compose the shared release branch PTC rename with the structural system-head migration in the existing V2-to-V3 stage. Audit source payloads and remap local coordinates before renaming dispatch vocabulary and owned plugin attribution; retain original message identities and generation-qualified references. Native V3 validation combines the same system and PTC lifecycle semantics through a private released-validation view.
Reuse the V3 row-admission owner in current JSONL scanning before recoverable-tail suppression, so malformed rows cannot hide retired required PTC tags or header.system. The new corruption-prefix regressions and all173 existing JSONL scanner tests pass. Combined migration/catalog/persistence suite:300 tests; executable V3 migration source has100% coverage. No released generation, prior migration implementation, shared release branch, benchmark or CI workflow is modified by this integration.
Regenerate advanced and restart expectations through the shipped macOS arm64 SEA runtime built from 0521815d45. Real delivery-accepted events now record sessionFormatVersion 3 in the five V3 logs and advanced SDK notifications. Restart result.json and requests.json regenerate identically. Preserve all five V2 predecessors byte-for-byte from b2b300f9; no production, migration, launcher, or test changes.
Evidence (macOS arm64): pnpm exec tsx scripts/build-exe-for-python-sdk.ts --targets node24-macos-arm64 passed, including the full pnpm build. file dist-exe/deepseek-harness-sdk-runtime-macos-arm64 reports Mach-O 64-bit executable arm64.
For each SCENARIO=sdk-snapshot,sdk-restart: PYTHONDONTWRITEBYTECODE=1 PYTHONPATH=python/sdk/src:python/sdk-runtime/src /Users/cty/dsh-deploy/.agents/worktrees/sp-kv-sdk-evidence/tmp/py-sdk-venv/bin/python scripts/smoke-python-runtime.py --exe dist-exe/deepseek-harness-sdk-runtime-macos-arm64 --scenario SCENARIO --update-snapshots passed; the same two commands without --update-snapshots passed.
PYTHONDONTWRITEBYTECODE=1 PYTHONPATH=python/sdk/src:python/sdk-runtime/src uv run --no-project --python /Users/cty/.local/share/uv/python/cpython-3.14.5-macos-aarch64-none/bin/python3 --with pytest --with pydantic>=2.12,<3 python -m pytest python/sdk/tests/test_smoke_model.py -q: 28 passed. Earlier helper invocations lacked pytest or SDK imports; corrected environment passed without source changes. git diff --quiet b2b300f9 -- scripts/snapshots/python-sdk-single-exe/**/*.v2.jsonl and git diff --cached --check passed.
Restore all 147 historical canonical fixtures changed by the lower branch to exact shared base 220ff708e3628a9be46d9747d09af6d8cd742f0d bytes (145 V2 and two V0). All 180 committed V0/V1/V2 canonical files now have zero diff against that base, including unchanged Python, test-support and preview sources.
Preserve corrected lower writer bodies in 146 V3 successors across 119 owner directories: replace 135 unreleased V3 counterparts and add 11 siblings. Change only each copied Session header version; preserve all body bytes, raw model streams, inline images, IDs, references and historical delivery generation markers. Remove two unshipped V2 additions after creating their V3 sibling: record-suite/rec-pin and empty-response-retry-current.
Keep six explicitly historical snapshot owners pinned with no V3 sibling. Restore the two pre-step V0 record child sources without synthesizing current output. Fix seven embedded Session header versions in exactly five test-support behavior owners whose output filenames already select V3. Leave protocol expectations, shared references, sidecars, consumer scripts and production migration/normalization code untouched.
Validation: purpose-built temporary Python inventory and verifier checked 337 canonical headers, 3714 successor JSON rows, message roles, contiguous child roles, exact successor bodies, all 180 historical base bytes and zero git diff, noncanonical JSONL immutability, six retained owners and six unchanged preview fixtures. git diff --cached --check passed. Temporary generators removed. Strict migration, native refresh and replay remain pending parent integration; unsupported pre-step sources are reported separately.
The representation PR left child and restart recordings in the old header-system representation while the packaged writer emits system/message before entered user messages. Restart result expectations also retained standalone assistant/chunk notifications after the writer moved stream records into assistant/message. These are stale expected artifacts, not fields to erase in normalization.
Regenerate the owning advanced and restart scenarios through the native macOS ARM64 packaged dsh runtime. Retain messages, tool effects, typed feedback and packed streams; update only the missing system nodes and their sequence references, and remove obsolete standalone chunk notifications. The advanced parent result and parent session already match the writer after the rebase, so this commit changes only two child logs and the restart result/logs.
Validation: pnpm run build; pnpm exec tsx scripts/build-exe-for-python-sdk.ts --skip-build --targets=node24-macos-arm64; uv run --project python/sdk python scripts/smoke-python-runtime.py --scenario sdk-snapshot --exe dist-exe/deepseek-harness-sdk-runtime-macos-arm64 and the equivalent sdk-restart command each reproduced the mismatch, then passed with --update-snapshots and again without it. Both read-only reruns pass. Other native targets remain covered by exact-head CI.
Consolidate the representation-change PR and its rebase reconciliations into one baseline. Preserve the exact tree and keep the in-history feature in the dependent PR. Follow-up fixes remain separate.
Windows x64 runtime builds resolve their hosted matrix.runner unconditionally
again (windows-2025 for pull-request CI). Remove the DSH_CI_FAILOVER_WINDOWS
selector, job-private Python toolchain, self-hosted dependency install and
post-step cleanup, the private setup script, and the routing spec introduced
in #3629. The Windows failover switch again covers only the native Windows
jobs in ci.yml.
scripts/test-proxy-environment.spec.ts pins the Vitest configs that declare
setupFiles so the proxy isolation setup is never dropped; the benchmark
lane declares it too.
Add a benchmark lane (`vitest.bench.config.ts`, `pnpm run test:bench`,
gate mode `ci-bench`) and a required `node 24 / benchmarks` CI job that
runs it alone. Benchmarks synthesize their input in-process from fixed
parameters and fail on documented budgets:
- `open-generation.bench.ts`: a 200-turn released-v0 log with 500 text
and 125 reasoning deltas per reply (127,400 events, ~2.8 MB) encoded
through the frozen v0 codec; the migrating first `open()` must finish
within 2,000 ms in a child process capped at 128 MB of old space, and a
fresh process must open the published current generation within 500 ms.
- `conversation-fold.bench.client.ts`: 200 replies whose compact streams
hold 2,000 text + 500 reasoning deltas each, folded through every Chat
Definition by the real assembler; the fold must finish within 150 ms
and stay within 3x the fold of the same window with 100 deltas per
reply.
On this commit both gates fail: the migration exhausts the 128 MB heap
(4.8 s and 696 MB peak RSS without the cap; the pre-stack decode of the
same bytes took 34 ms and 168 MB) and the fold scales 11x with the delta
count. The stacked fixes bring both paths to O(records).
The probe's child was a script string that wrote its report to stdout, and
the parent parsed the whole stream as one JSON value: a plugin that merely
printed a line at import could not be probed, and a package that kept a
timer alive after import ran until the timeout. The parent asserted the
parsed value as the report, and the cache read checked only `format`, so a
file holding nothing but `{ "format": 2 }` came back as a complete record.
The child is now its own module, `probe-child.ts`, run through tsx under a
source launch and bundled as `lib/probe-child.js`, and it sends one report
over an IPC channel; stdout is not read at all, and the child is killed as
soon as the report arrived. `parseChildReport` and `parseProbeRecord`
validate the report and the cached record field by field, so an
unrecognized report is a rejection and an unrecognized record is probed
again. `PluginProbe.ok` now states what it is: the main export imported and
cordis is not a second copy; `kind` and `addable[].ok` decide what can be
enabled or added.
Tianyi Cui