The meter regression deliberately rewrites the system node during request-error recovery. After the retry reconciliation fix, the next request must restore the captured admitted prompt rather than trust that temporary replacement. Keep the replacement stimulus and zero-delta metering assertions, but compare the retried prompt with the initial request so the two independently corrected components are tested together.
Validation: the combined regression run exposed this obsolete retry-text expectation while 945 other cases passed; all eight compaction-loop-repro cases pass with the admitted-prompt assertion.
The loop appends step/start before system/message and the entered user
messages. Capturing nodes at step/start therefore omits inputs already
included in the provider's successful usage, then adds those inputs back
as a positive surface delta. Prompt replacement can also incorrectly add
or subtract the difference from the prior prompt on a completed call.
Snapshot the current priced surface immediately before assistant/message
commits. Keep provider output separate from the durable assistant node so
listener rewrites retain their signed delta. The invariant is zero delta
immediately after an unchanged successful output: provider usage already
includes every admitted prompt input. Later appends/replacements still
produce signed deltas, and low or absent usage keeps heuristic fallback.
Delete stepStart.nodes rather than adding prompt-specific corrections or
another request snapshot: the existing transactional surface fold already
contains the successful request inputs, including replacements made during
same-step retry recovery. Keep turn/step state and all overlap, mismatch,
and late-assistant lifecycle validation. Retry attempts are log-only and
request middleware changes configuration; injected messages remain queued
until admission. No loop, event format, projection, or retry policy changes.
Exercise the real loop with reported usage and initial, growing, shrinking,
and empty prompts; same-step failed attempt plus retry prompt replacement;
request middleware; eager observation and fresh seeded replay. The two
regressions fail before the fix with spurious deltas of +48 and +18 tokens.
Retain existing durable-output rewrite, route repricing, missing/low usage,
transactional failure, and lifecycle tests. Update README EN/ZH and the
existing system-prompt surface-node Agent Note, including pairing records.
Validation (dedicated worktree, no full unit suite):
- pnpm exec vitest run packages/llm/token-meter/tests packages/compaction/compaction-basic/tests/compaction-loop-repro.spec.ts --coverage --coverage.include='packages/llm/token-meter/src/index.ts'
118 passed; exact changed runtime file 100% statements/branches/functions/lines.
- pnpm exec vitest run packages/core/agent-loop/tests/request-reconstruction.spec.ts packages/compaction/compaction-basic/tests/compaction-basic.spec.ts packages/compaction/compaction-basic/tests/loader-composition.spec.ts
118 passed, including retry reconstruction and real Loader composition.
- pnpm run doc-sync: 33 gates passed.
- pnpm run test:docs: 15 gates passed.
- pnpm run lint: passed, 0 warnings/errors.
- git diff --cached --check: passed.
Baseline normalize.ts comment conflict marker is intentionally untouched.
The session already derives the protected system head as a Message, and both adapters accept leading system history. Passing it through a separate SummarizationInput.system string unnecessarily flattens that value and rebuilds the same wire message in the adapter. Prepend the derived head to messages and remove textContent, the separate field, and GenerateOptions.system plumbing from the summarizer.
Keep range selection, shadowed seq accounting, session head protection, routed tools, image references, target policy, and the model-visible compaction instruction unchanged. Empty-content heads still derive to null and contribute no request message, but their surface node remains protected. Update subclass consumers/tests, EN/ZH package and subsystem prose, and the existing system-prompt surface owning note with refreshed pairing records.
Evidence: pnpm exec vitest run packages/compaction/compaction-basic/tests packages/llm/llm-deepseek/tests/serialize.spec.ts packages/llm/llm-pi-ai/tests/context.spec.ts --coverage --coverage.include='packages/compaction/compaction-basic/src/region.ts' --coverage.include='packages/compaction/compaction-basic/src/summarizer.ts' passed 203 tests in 6 files; both changed sources have 100% statements, branches, functions, and lines. Region-to-default-summarizer cases pin exact prefix and tools for nonempty Unicode/multiline, empty, and absent heads. DeepSeek JSON byte equality and pi-ai context equality pin leading-message vs separate-system equivalence on text and image-capable conversion paths.
pnpm run doc-sync passed all 33 gates including doc-typecheck, documentation build, translation pairing and model-experience checks. git diff --check passed. Own dependencies installed with pnpm install --frozen-lockfile. An initial test iteration used a nonexistent ctx.dispose teardown on the in-memory fixture; corrected to its existing fixture lifecycle and reran successfully. No runtime/model behavior, normalizer marker, main worktree, push, or rebase changes.
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.
The persistence seam is now create/open/stat/list returning per-session
SessionHandles (read/append/flush/close); every log read and write flows
through the owning handle. The seam package exports only the service and
handle contracts, consumer-visible errors, and pure durable-data
validation helpers; each backend owns its complete storage runtime, and
the shared contract suites pin equivalent observable behavior. The
backend routes published sessions' live events by id into the active
write handle; agent-loop only acquires, seeds, and closes the handle.
Resume appends interruptedTurnClosers through its write handle;
session-query owns the revision-keyed cold cache. Legacy-only surfaces
are removed in the same swap: locate/readRaw/supportsRawArtifacts, the
legacy event-shape read migration, zstd torn-frame salvage,
DSH_SESSION_JSONL, and hook transcript_path population; a torn final
zstd frame is discarded whole; the session-list cold blank probe returns
on stat metadata (eventCount derived from the last physical row,
sizeBytes). The WebUI ZIP export serializes the logical log from a read
handle, so both backends export identically.
Refs #3245
Tianyi Cui