Repository navigation
Conversation
IcyDate
force-pushed
the
2026-10-27_durable-agents-md-update-carriers
branch
from
October 27, 2026 05:32
d6ea911 to
0d6578a
Compare
Owner
(ai: vllm-qwen/qwen3.8-27b) 查重：不重复。 open+closed 全量检索（按 标签：bug + dsh + scope/kernel + p2（缺陷类；dsh 车道；src 侧折叠保护判定属压缩决策面；条件触发、有绕行，p2）。 本条为外部贡献 PR（无 agent 标记），正在做完整审核：拉分支读 diff、核 CI、定向跑相关测试。结论稍后按八问体例补在这条线程。 |
Owner
(ai: vllm-qwen/qwen3.8-27b) 审核完成（外部贡献 PR，独立审过 diff、机制与 CI）。结论：可合。 ① 方向关：修对了层。缺陷的产地是分类器—— ② 回归审查（逐项过既有行为）：
③ 代码审查（主关）：关键机制我独立核过源码而非只信 PR 描述—— ④ 测试有效性：新增 e2e 折叠场景走完整管道（ ⑤/⑤b diff 整洁度与代码质量（外部提交从严）：3 个文件全部与目的相关，无空行/生成物/锁文件噪音，无 version 触碰；命名达意（ 八问：
收工判定：代码与行为均可合入。本 PR 改变默认保护行为面（§7.4「fallback/default 变化属产品决策」），不进 auto-merge，请人工合并：#4922 。合并后 #4921 由 PR 的 |
Owner
[bot] 🏷 (ai: vllm-qwen/qwen3.8-27b) |
Owner
[bot] 🏷 (ai: vllm-qwen/qwen3.8-27b) |
… issue number Repo convention names regression files after the GitHub issue (issue2915-/issue3236-/...); GitHub ranxianglei#5767 is an unrelated banner issue, so the mirror number in these filenames pointed at the wrong thread.
…IdArg The ad-hoc id normalization kept original case and skipped zero-pad forms, so B5/005-style ids undercounted the freed-chars receipt; parseBlockIdArg is the kernel's canonical normalizer and unknown ids are skipped honestly instead of matched against a mangled key.
acp_retire/acp_restore/acp_rewrite fell through toolChipCls to the unstyled default; they sit in the same fold-family chip class as compress/acp_status/acp_cache.
…#5765 review) acp_retire counted freed summary chars over the whole input list before the flip — already-retired (noop) and rejected ids inflated the receipt. Count outcome.applied after the flip instead (summary text survives the flag change). acp_rewrite silently dropped malformed items[] entries — the model could not tell which item did not land; report a skip count. Drop the dead failedCount counter.
…cks (ranxianglei#5765 review) Math.min(...nums)/Math.max(...nums) over effectiveMessageIds can exceed V8's argument-spread limit for tier-3 blocks of very long sessions; compute min/max with a loop. Reuse the already-computed active list in renderHandoff instead of re-filtering s.state.blocks.
…eview) The PR's snapshot-render change (retired blocks contribute no synthetic anchor while covered rows stay suppressed) had no test anywhere; legacy prune and D-lane behave identically for this shape, so add a hermetic D-on test with assertDActive to prove the lane rendered.
ranxianglei#5481) The merge of origin/master into the branch (4954a12) resolved the CHANGELOG conflict with the pre-review wording of the ranxianglei#5481 entry, silently dropping the correction from fe4ffdb: the codex local-forge path logs its cost line WITHOUT an outbound field now (nothing is handed to fetch there), so '本地 forge ... 既有记账边界不变' was inaccurate for that path. Re-applies the corrected sentence verbatim.
darwin's 'security find-certificate -a -p' keychain export includes non-CA leaf identities (e.g. com.apple.kerberos.kdc, which also carries a duplicate EKU extension in violation of RFC 5280 4.2). Strict replace-semantics clients such as codex's rustls HTTP stack parse every entry of SSL_CERT_FILE and refuse to build their client when any entry fails, so a single such cert broke every connection through bili codex. writeCombinedBundle now keeps only entries that parse as X509 certificates AND carry basicConstraints CA:TRUE or the keyCertSign key-usage bit (legacy pre-basicConstraints anchors), applied to ALL sources (system bundle, OS store, Mozilla roots, MITM root). The bundle gains a leading marker comment; a fresh file missing it (pre-upgrade bundle) is rebuilt once so the fix takes effect immediately instead of waiting out the 24 h freshness gate. If the filter ever drops everything (broken certificate parsing), it falls back to the unfiltered bundle with a loud error log rather than shipping an empty trust store.
…L minting arm Owner ruling (ranxianglei#5608 decision thread follow-up): the three one-shot full-prefix-invalidation producers get fold-boundary discipline, closed inline on PR ranxianglei#5609 per 「不要立issue，原来的pr改」. C — timestampsHighWaterRefSeq: a rendered ACP tag gains its time= attribute only when its ref index exceeds the mark recorded pre-pass. The mark freezes from state across sync/fold/hydration and ledger alignment raises it to the aligned max — so flipping ranxianglei#4313 on mid-session (incl. the v0.1.209→v0.1.210 default off→on upgrade) changes ZERO prefix bytes until a tag's own content edit/fold remints it onto a higher seat. Brand-new sessions stay fully functional. B — suppressMintingOnce: the stamp node records ids without fabricating Date.now() while armed, instead of seeding false timestamps that would ride into future lineage via seedTimestampsFromParent. Host arms it on bulk-old-history paths (miss-adoption fallbacks, weak adoption, forks); strong byte-exact resume stays un-armed (uncovered tail provably new). Self-clearing; deliberately not hydrated after restart. kernel/package.json 0.0.180 → 0.0.181. New pins: kernel/tests/issue5608-timestamp-tails.test.ts (4).
…the next post-fold save B/C/D ruling 「不能留尾巴」(D half): a restarted proxy must serve bytes identical to what the provider cached — renumbering every historical ref at cold load invalidated whole prefixes on every restart. Restamp now fires once from the persistence path after a fold (pendingLedgerAlign marker set at the stream.ts fold commit, consumed by scheduleSave / flushSync / writeNow before content-store serialization so CCR re-keying (ranxianglei#4733) rides the same record). Sessions that never fold keep legacy counter refs forever at zero cost; stale-citation refusal (ranxianglei#4599) covers the interval either way. Also arms honest-NULL (kernel ranxianglei#5608-B) on the miss/fork boundary paths that bulk-seed OLD history into fresh child state (handle.ts fallbacks, weak adoption, derived/manual forks); strong byte-exact resume keeps honest now-mints for its provably-new tail. Re-pins S15, the HTTP fork survival test (cold load is now a byte-exact no-op) and the storage digest pins; the miss-seed e2e pins replace fabricated-now assertions with the honest-NULL contract.
…glei#5609) Only conflict: kernel/package.json version line — both sides bumped after base 0.0.180 (ours 0.0.181 via ranxianglei#5608-B/C time-attr work, master 0.0.182 via ranxianglei#5758/ranxianglei#5759). The merged tree carries BOTH unreleased change sets, which no single prior version denotes, so reconcile to 0.0.183 (max+1).
…#5802) Review finding on PR ranxianglei#5809: the raw-DER KeyUsage scan expected the extnValue OCTET STRING immediately after OID 2.5.29.15, but per RFC 5280 s4.1 a critical-marked extension carries a BOOLEAN between the two. KeyUsage is RFC-recommended critical on CA certificates, so exactly the legacy pre-basicConstraints anchors the fallback branch exists for were silently dropped whenever they carried the marking. Empirically verified: a forged keyCertSign-only cert with the critical flag parsed isCaLike=false before this change, true after. Adds the matching regression row with ground-truth assertions.
… states Conditional spreads in the two kernel clone paths (sync.ts syncBlocks clone, compress.ts fold clone) plus delete-style flag clearing in the stamp node: states created before this feature must not acquire the `timestampsHighWaterRefSeq` / `suppressMintingOnce` KEYS with undefined values — key presence itself moved golden stateKeyDigests and would have flapped parity digests turn-to-turn (golden-storage-render suite red on 3edf28b). The B arm still clears the key (not `undefined`) so a consumed flag leaves no trace; C writes its mark as a plain number after first init. Version 0.0.183 -> 0.0.184.
ranxianglei#554 state poisoning) The first cut armed suppressMintingOnce on every "starting fresh" fallback of the resume-inheritance gate. That is the same false-positive zone v3's gate already rejects adoption for: cross-lane boilerplate prefixes earn positive overlap against genuinely NEW conversations too, and lanes that never run a kernel pass (side requests are pure passthrough, ranxianglei#388) can never self-clear the flag — the arm leaked into persisted kernel state and broke the ranxianglei#554 side-request state deep-equals in CI. Decision: unproven identity keeps brand-new semantics (its ids mint normally); arms now live only on proven-identity paths — weak/strong resume adoptions (maybeAdoptResume, armBoundaryNull) and derived-fork/manual-fork sites. Below-threshold restarts therefore mint "today" display values on uncovered lines again: display-only cost, accepted; billing behavior is unchanged either way. Re-pins: - tests/side-request-isolation.test.ts green again (no fixture change) - tests/resume-miss-timestamp-seed.test.ts gate-negative e2e: FRESH fallbacks must MINT their own fresh ledger (was: empty/honest-NULL) - tests/storage-orphan-recovery.test.ts: a bare cold load arms nothing (marker is fold-commit-only, ranxianglei#5608-D) - src/fork-adoption.ts / src/plugin.ts / src/state-hydration.ts: guard the timestampsHighWaterRefSeq inherit/carry so a markless parent cannot materialize the key on child/persisted state
…ndle-ca-only-filter [PR] fix: keep only CA certificates in combined-ca.pem (ranxianglei#5802)
…route-env-on-unload [PR] fix(hermes-plugin): restore routed process env on unload and watchdog exit (ranxianglei#4644 follow-up)
kernel version: 0.0.183 (PR) vs 0.0.192 (master) -> 0.0.193
…anxianglei#6132) Owner-directed full removal of the experimental compressAtFoldPoint / compressAtFoldPointClearThinking feature (v1 ranxianglei#5471 + unreleased v2 ranxianglei#5949): two distinct live pathologies in 9 days on its own target model family — 2026-11-08 voluntary re-compress loops (v1, ranxianglei#5949 field report) and 2026-11-10 authorship misattribution (v1 in v0.1.213 production: the pair's relocation left the turn's own answer text bare at the transcript tail, and a 27B model read its own last utterance as a user message, m00376 chain). - delete kernel/src/relocate-compress.ts + kernel/tests/relocate-compress.test.ts - drop the relocate-compress-carriers pipeline node (after hide-compress-calls, before view-latch) and its registration - drop Config.compressAtFoldPoint / .compressAtFoldPointClearThinking / .reasoningDroppable and the public relocateCompressCarriers/plan exports - drop the already-folded-echo parse kind + v2 pointer marker (parse-compress-input.ts) and the two echo tests - kernel 0.0.192 -> 0.0.193 (both package-locks) kernel-change: removed relocate-compress.ts, the pipeline node, 3 Config fields and the public exports (full feature removal, owner decision after the 2026-11-10 incident, ranxianglei#6132); with the switch off the transform was a byte-identical no-op, so removal changes nothing for the default layout. kernel 0.0.192 -> 0.0.193. Closes ranxianglei#6132 (kernel side; host/web/docs follow in the next commit of this PR)
…#6132) Companion to the kernel removal: strip the experimental fold-point relocation feature (v1 ranxianglei#5471 + unreleased v2 ranxianglei#5949) from everything else. - env knobs BILI_COMPRESS_AT_FOLD_POINT(_CLEAR_THINKING) + resolvers (knobs.ts), file keys compress.compressAtFoldPoint(_ClearThinking) (config.ts: CompressSettings fields, COMPRESS_SETTING_FIELDS, parser loop), compress-settings.ts cascade/pass-through/reachability, relay.ts round-2 mirror + drop-predicate install, the four prepare* install sites, plugin-render.ts S4 mirror, makeFoldPointReasoningDroppable (server.ts), the v2 already-folded-echo ack in stream.ts - stale configs: one-shot INERT deprecation warning extended to the two removed keys (every level incl. top-level), values ignored - web config row + i18n strings (en/zh), config-reference yaml, site docs, site i18n strings, blog post + article entries deleted (owner instruction) - docs configuration en/zh + lifecycle ledger (rows removed, §5 note), CHANGELOG entry - tests: delete tests/compress-at-fold-point.test.ts, drop the echo test from compress-reject-reasons.test.ts, extend removed-compress-keys-inert with the removed fold-point keys Closes ranxianglei#6132
…5765 tool surface The acp_retire/acp_restore/acp_rewrite schemas are always-on in every plugin manifest, so the pinned manifest/fork/launcher tool lists, the insertion-order lists, and the codex-official prompt regex ('four' -> 'seven' context-management tools) all needed the three new names. Fixes the 9 stale-assertion failures across plugin-protocol, plugin-fork, launcher-plugin-mode and codex-official (deterministic on PR head, green on master).
… surface The four committed storage-render snapshots are byte digests of the forwarded wire bodies; the always-on retire/restore/rewrite tool schemas change every body (+2685B on the chat wire per turn), so the digest/bytes fields were refreshed via scripts/update-storage-render-goldens.ts. Structural fields (turns, summary placement) are unchanged. Fixes golden-storage-render + storage-engine-proofs (8 tests, 4 wires).
…y growth The always-on retire/restore/rewrite tool schemas add ~0.66-0.7k est-tokens (~2.6KB) to every forwarded body. Several e2e scenarios are deliberately calibrated a few hundred tokens under their trigger boundaries (the repo's established pattern when the wire surface shifts: ranxianglei#2750 wording change, ranxianglei#3298 fold-boundary retune, ranxianglei#1933 bill-rate retune), and the constant body growth pushed them over: - e2e-relay-5xx-selfheal: tool_result 12000 -> 11700 chars keeps the payload at the intended ~97% of the 32k window. - issue1933 F2: 1.4x/15x4000 -> 1.5x/14x4000 keeps the billed total at exactly 21000 (the scenario's +1k-above-target) while the estimate lands ~1.5k under the 20k trigger. - ranxianglei#663: repeat(235) -> repeat(220) keeps the post-fold forward at 4 messages (the mock only misclassifies exactly-1-message bodies). - e2e-google-compress: clamp the mock's fold endId to refIds[max(len-6,4)] — with the +2.6KB body the 24KB demand gate now opens at 12 refs, and the len-10 tail then degenerates to a single-message range (463 chars < minCompressRange, hard refusal). The len-6 clamp also fixes a steady-state ratchet each fold out-earned by ~500B, so the upstream bodies stay bounded (client 65.7KB vs peak 31.5KB, ratio 2.09, assertion requires >2). - ranxianglei#280: repeat(235) -> repeat(155) — the overflow-relaxed preflight fold now eats 3 more messages before the hard protection floor; 155 (the measured pass window is 100..215) keeps MARKER_8..11 in the forward. - ranxianglei#1843: mock bill 8000/6617 -> 8657/7274 (a real encoder bills the +657-tok schema text too) so the learned per-image cost lands at 1813.5, inside the (1500, 5000) sanity bound. - preflight-hold ranxianglei#568: repeat(235) -> repeat(220) (same family as ranxianglei#663). - issue2672: idx8 2562 -> 207 chars (it is BLOCKED — no compressible slot — so its size never touches the walk geometry) and tail 358 -> 294 chars (74 tok each keeps the 1000-tok recent-zone walk stopped at idx9 exactly as on master: 5x74+5+675 = 1050 >= 1000). Restores the pre-ranxianglei#5765 post-fold payload (~7.6k tok vs the 8k window). - cache-proof host-compaction: seed FILLER 6 -> 5.5KB/turn — the +2.6KB schemas pushed the second-phase seed body one body earlier past the 64KB demand threshold (66.9KB vs 65.5KB at body ranxianglei#11), landing the seed fold one turn ahead of the ranxianglei#2695 geometry the zombie-settle bound (9) and the rearm-shrink pin were measured against; 5.5KB restores the master-aligned fold turns (verified via PROOF_DUMP: seed folds at 8->9 and 12->13 on both, rearm at 21->22). All 15 affected files green (156/156), matching the 27 deterministic CI failures 1:1; the identical suite is 155/0 on master.
…nxianglei#5765) The ranxianglei#5924 architecture-docs guard (landed on master via the branch sync) requires every tracked src/ file in reference/architecture.md; the new block-ops host executor module was added by this PR without its map row, so the guard failed on the merged head. One-line registration next to the acp-* cluster; guard passes locally (320 src files covered, no ghosts).
…+ tag-stripped body (ranxianglei#6160) The scripted queue key was conv|sourceText.slice(0,64) over the RAW last-user text, whose ACP render tag is volatile by design: the post-fold ledger alignment (ranxianglei#5608-D defers restampRefsToLedger to the next successful fold) renumbers old refs (m00001 -> m00015, refs-never-reused) and raises the ranxianglei#5608-C time-attribute high-water mark, so the very next request re-renders the same message with time= dropped. The 64-char key slice contains the tag, so one scripted turn split into two queues the instant a fold landed and the directive queue replayed from scratch — the ranxianglei#2791 tests' deterministic "compress must be CALLED exactly once (got 2)" on local node-25 and the intermittent red on CI (same SHA red and green across runs; the alignment no-ops when the unified store has no visible rows yet at the post-fold scheduleSave). The key is now conv|<source ref>|<stripAcps(source).slice(0,64)>: the ref names the message occurrence (stable for the turn; a remint voids the script anyway) and keeps byte-identical prompts distinct (switch/back both send 请调用probe_tool); stripAcps (existing helper) drops the volatile wrapper. One site (answerFor) covers all six suites sharing the fake. Verified locally (pi-stable 0.83.6): both ranxianglei#2791 tests green — the fold run's queue continues through the fold (probe -> compress -> acp_status, one call); full pass over native-pi/dsh/opencode/delegate-notify/subagent-sessions (the two remaining local failures reproduce identically on the unmodified fake: pre-existing local-environment artifacts). Closes ranxianglei#6160
…e-compress-at-fold-point [PR] remove: the experimental compressAtFoldPoint feature — fold-point relocation (v1 ranxianglei#5471 + unreleased v2 ranxianglei#5949), ranxianglei#6132
…ght + role cards) The sessions' handoff (交接文档) tab rendered every message body as escaped plain text: markdown in assistant replies stayed literal, code fences showed raw backticks, and tool input/output was one giant unhighlighted pre with no copy button. - new src/web/mdrender.ts: single escape-first Markdown + hand-rolled syntax-highlight core (json/diff/js/ts/py/sh lexers, GitHub light+dark palettes, unknown languages degrade to escaped text, zero runtime deps). Inlined into the browser IIFE via .toString() exactly like the JSONC helpers; tests/mdrender.test.ts pins behavior plus the self-contained inlining contract (esbuild-minified .toString() must run in a bare scope, mirroring the tsup pipeline). - web/markdown.ts becomes a thin wrapper over the same core, so the server-rendered handoff fallback and blog articles gain highlighting too. - client.ts renderHandoffMd: role cards (user/assistant/tool left border + badge), markdown-rendered bodies incl. thinking chains (collapsed by default), structured tool cards (input/output label, language detection json/diff, per-block copy button, JSON pretty-print, open when <= 20 lines), document meta (title/session id/token stats) rendered as a key/value box instead of being skipped. - styles.ts: card/token CSS vars in light + both dark palettes (ranxianglei#2559 verbatim-equality kept); i18n: det.io_in/det.io_out. - tests/web-handoff-roles.test.ts adapted to the new markup; the ranxianglei#2065 invariant (in-body headings never become role dividers) is preserved. CHANGELOG + docs/architecture/webui(.zh-CN).md + reference/architecture.md file map updated (architecture-docs lane green).
…ei#5984) dsh-live-compact is a competing compaction engine that takes over dsh's stock compaction row (its bundled cordis.patch.yml disables compaction-basic and registers its own engine row; its preset takeover renames the preset's compaction row to itself), so bili's load-time compaction-disable overlay (ranxianglei#3354), which recognizes only the stock row names, is structurally blind to it. Its checkpoints can land outside bili's knowledge and destroy the compression substrate - the ranxianglei#5984 session. - scanDsh: dsh-live-compact entries (bare name / npm spec / bounded path segment, isDshLiveCompactEntry mirroring isOpencodeAcpEntry's discipline) are now a deterministic "known" finding (knownId "dsh-live-compact") instead of a keyword [suspected] one, so the conflict ledger / acp_status / web banner surface it as a confirmed conflict; lookalikes stay at most keyword-suspected - announcements/remote.json: web notice (ranxianglei#5548) for dsh profiles with the plugin, with the removal recipe (dsh plugin remove dsh-live-compact) and a CTA to ranxianglei#5984 - docs: known-conflicts list updated (configuration.md / zh-CN) - tests: 3 new cases + detail-shape / not-benign / not-sibling / not-display-only assertions
…ts own section The client renderer only recognized ### user/assistant/tool as block dividers (ranxianglei#2065), so the export.ts ranxianglei#845 appendix ('## Compressed block summaries' + one '### Block bN' entry per active block) landed inside the last assistant card — on a session with 72 active blocks that was ~700 lines swallowed into one card. '## ' headings after the Conversation heading now close the role cards and open a section; '### Block bN — topic' entries render as small cards (mono block id + topic + tier/token meta line + Markdown body). ranxianglei#2065 invariant intact: in-body '### 7.1 …' lines are still content.
….blockOps (default off)
Owner ruling (review of this PR): acp_retire / acp_restore / acp_rewrite
must sit behind ONE switch, default OFF, injected only when ON — the
always-on surface costs ~0.7k estimated tokens per request on every
session of an enabled proxy.
Follows the absorb opt-in pattern end to end:
- kernel: definitions + name consts stay exported, but the 3 tools leave
the default arrays (ACP_TOOLS_{OPENAI,ANTHROPIC,RESPONSES,GOOGLE},
ACP_READONLY_TOOLS_RESPONSES, ACP_TOOL_NAMES); they stay in
ACP_MUTATING_TOOLS (dispatch must drive the compress loop when enabled).
New BLOCK_OP_TOOL_NAMES set + buildBlockOpsSystemPrompt(lane) section
builder; the default prompts are back to "four context-management
tools". New Config.blockOps { enabled } (off by default); absorb/ccr
tool-name predicates include the block ops.
- host: compress.blockOps config (3-level merge like absorb);
blockOpsEnabled / effectiveBlockOpsConfig / storeEffectiveBlockOps
(per-session, plugin tool API parity); isProxyToolFor, the proxy-lane
inject* extras, the responses hybrid lane, the round-2 cache seam, the
plugin manifest + tool API, the MCP dialect note and the system prompt
all inject/advertise the 3 only while enabled; the MCP note, prompt
and per-session metadata stay byte-identical to pre-ranxianglei#5765 while off
(zero footprint: no metadata key written while off — the storage-render
goldens match master byte-for-byte).
- dispatch: enabled calls run the kernel ops exactly as before; a stray
call while disabled gets a model-facing "not enabled on this bili proxy
(compress.blockOps.enabled is not true)" receipt and never touches state.
- wire-contract goldens revert to the master shapes (the default surface
is unchanged), and this commit supersedes the three earlier fix commits
on this branch (test-pin updates + storage-render regen + payload
retunes): with the default surface back to the master shape, those
test files return to the master versions and the 27 red CI tests are
green again without any calibration changes.
- docs: compress.blockOps in configuration.md/.zh-CN.md + website seed;
kernel 0.0.194.
Verified: kernel 1656/1656; host targeted 201/201 (the 15 previously
red CI files + block-ops/snapshot/wire-contract), 348/348 + 63/63
surface/config invariant files; typecheck + build green.
…el 0.0.195) Conflicts: kernel version files (master 0.0.193 from the ranxianglei#6132 fold-point removal + this branch's 0.0.194 -> 0.0.195) and one relay.ts import line (master dropped relocateCompressCarriers; this branch adds buildBlockOpsSystemPrompt — the resolved import carries only the latter; no fold-point call sites survive in the merged relay). Re-verified on the merged tree: kernel 1631/1631, host targeted 201/201 + surface/config 374/374, typecheck + build green.
… mcp-note split unused-exports guard: MCP_DIALECT_NOTE_BLOCK_OPS and mcpDialectNoteText are consumed only inside compress-tool.ts (withMcpDialectNote) — make them module-local.
… gate - lock-sync: the second master merge auto-merged the root package-lock with the pre-merge kernel version (0.0.193); regenerate it for 0.0.195 (npm install --package-lock-only). - lane-guard: src/server/side-request.ts (BILI_TOOL_NAMES += the kernel- fixed block-op names) is side-lane logic — pin it in the dedicated side-lane suite: tests/side-request-isolation.test.ts now asserts the bili-owned set covers ACP_TOOL_NAMES + block ops + absorb/rule/retrieve.
…ive-compact-known-conflict [PR] thirdparty-scan: flag dsh-live-compact as a known conflict (ranxianglei#5984)
…for pruned leaf messages Two fixes for the session-detail 压缩结构 tab (ranxianglei#3917 tree): 1. Node titles flickered away on select/poll redraw because they were native SVG <title> elements. Replaced with a persistent in-page floating tip (.g-tip): follows the pointer, shows block id/tier/ message span/full topic/tokens/message count, and survives node selection and the 5s live-tree redraws (re-anchored after redraw). The <g> keeps an aria-label for accessibility. 2. Clicking a leaf block listed its message refs (mNNNNN) but every row said 'no longer in the session' — host native compaction prunes the messages from context, so the bytes were nowhere to look. They ARE on disk though: collectBlockContent froze each original message ([role] / [role • tool] header + text, pi-host user messages carry their ACP ref marker) into the block's full snapshot. New frozen-snapshot hydration: for a LEAF block whose full snapshot parsed fragment count equals its direct-message ref count — with every ACP marker cross-checking its own ledger position — the rows re-align by position (preview + click-to-open via /message?ref=). Count mismatch, marker conflict, or non-leaf spans (children mixed into the snapshot, position meaningless) abort the whole block and keep the honest 'missing' marker — never serve shifted bytes. Measured on a 253-block/9771-ref session: 5072 (52%) leaf messages recovered their real previews and text; the rest stay honest misses. Regressions: tests/web-blocks-tree.test.ts (2 new cases: positional hydration + the three refusal arms), web-sessions/web-handoff-roles/ mdrender and the web sandbox suites stay green; typecheck + check-dist-external (zero runtime deps) pass.
…emon [PR] feat: compress-block retirement (acp_retire/acp_restore) + external summary rewrite (acp_rewrite) + export ledger
…snapshot hydration - compression tree: hovering a node now widens its topic line in place (~20 chars) and the .g-tip overlay follows the pointer; a 5s live-tree tick previously redrew the graph with lean nodes and degraded every topic back to a bare block id — ticks now merge onto the rich detail blocks by blockId before redrawing (structure/active/spans win) - snapshot hydration gains a restamp-proof ARITHMETIC anchor: the marker's frozen number minus the block's frozen span start equals the snapshot position. Cold-load restamping (ranxianglei#5196 parked bands take fresh numbers) renumbers whole blocks, which invalidated the identity proof (marker number no longer in the ref list) and left entire renumbered blocks (b27: 127/127 missing) at honest-misses although their frozen snapshot held every byte. Identity still wins where the marker number is current; a marker proving neither aborts the block; unbalanced segments stay missing. Measured on the 253-block/9771-ref session: 7789 refs (80%) now hydrate (was 5072/52%), b27 fully recovered - tests/web-blocks-tree.test.ts: +3 hydration cases (restamped block recovered / arithmetic beats a coincidental ref collision / unanchorable marker rejects the whole block incl. click-to-open); marker-shift fixture span pinned so the case stays legacy-anchored
…-write-lifecycle [PR] fix: await unified snapshot writes and reset each debounce batch
…opic-trailing-system-tails [PR] fix: insert Anthropic ephemeral tails before trailing client systems (ranxianglei#5432)
…work [PR] Web Panel: Handoff Documentation and Collapsed Tree Diagram Rendering Upgrade (eWork-Level Rendering, Appendix Sectionin | Web 面板：交接文档与压缩树图渲染升级（eWork 级渲染、附录分节、冻结快照水合）
…able-message-guards.ts and issue5142-expected-rebase.test.ts) The two conflicts were in doc comments and a test assertion that both sides had updated independently: - master added image-handle/image-offload classifiers (12 total, 14 markers) - PR added workspace-instructions revision markers (4 total for that family) - merged: 12 classifiers, 17 markers Closes none — this is a CI sync commit for PR ranxianglei#4922.
Owner
[bot] (ai: vllm-qwen/qwen3.8-27b) 冲突已解决，PR 恢复可合并状态。 做了什么：把 master 最新 tip（
验证：typecheck 干净； 当前状态：head |
This branch has not been deployed
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
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.
这是啥问题
dsh 的 workspace instructions 是 durable 载体（#2447），而它有两种形态；守卫只覆盖了第一种。
基线是
Instructions from: <path>（大写 I），已被durable/agents-md保护住。但会话中途改了文件之后，@deepseek-ai/dsh-agent-instructions发的是核对消息：Updated instructions from: <path>Additional instructions from: <path>Instructions removed: <path>三者全是小写
instructions（或干脆换了动词），而dshWorkspaceInstructionsPath用的是msg.text.includes("Instructions from:")——大小写敏感精确匹配，一条都没被保护。后果不是"少了一条历史"。
baselineIdentity（dsh-agent-instructions/lib/index.js:39）只含projectRoot/projectRootMarkers/maxBytes/maxSourceBytes/instructionFileCandidates/localInstructionFileCandidates，不含文件内容——内容改了 identity 不变，宿主不重发基线。所以那条修订是本次变更唯一的载体。折掉它之后，上下文里留下的是「一条被钉死的旧基线 + 一个不含新规则正文的摘要」：模型继续按用户已经改掉的规则干活，并且无从察觉自己丢了更新。这比两条都折更糟——两条都折，模型至少知道自己手上没有规则。实测（一个 2907 条消息 / 23 个 block 的真实会话）：11 条
Updated instructions from:载体，9 条被折，其中一次compress一口气吞掉 4 条。另一会话里Additional instructions from: neighbor-help-frontend/AGENTS.md（6710 字符）被折进b1，此后整个会话再无项目指令。根因(代码坐标)
src/durable-message-guards.tsdshWorkspaceInstructionsPath：只匹配大写Instructions from:。原注释写着"大小写敏感是故意的——核对消息属于普通可消费历史"，本仓tests/issue2447-dsh-workspace-instructions.test.ts:138也把这一点钉成了预期。dsh-agent-instructions/lib/index.js:39的baselineIdentity不含文件内容，而:262/:283/:285三种修订文案全是小写（或换动词）——于是"可消费"的那条恰好是唯一带新内容的那条。toLowerCase()：默认策略是protectedLatestTools: ["durable/*"]的 latest-per-identity（collectLatestProtected按 virtualPath 分组、每组只留最新一条）。基线与修订共用durable/agents-md时，最新修订会把多文件基线挤成可折。实测那对 base + revision：基线带 2 个文件（全局 + 项目根，5311 字符）、Additional只带 1 个（6710 字符），共用 identity 就是从"保 2 丢 1"变成"保 1 丢 2"，净亏。改动
src/durable-message-guards.ts：新增agentsMdRevisionPath()，把三种修订投射到独立 identitydurable/agents-md-update/<path>；基线保留durable/agents-md。durable/*仍匹配带后缀的路径（matchToolPath对<prefix>/*只看第一段）。dshWorkspaceInstructionsPath改为先认修订、再认基线。这不是形式上的整齐：修订正文完全可能引用大写Instructions from:字面量（用户在自己 AGENTS.md 里说明这个匹配串），按旧顺序会被误判成基线、拿到和真基线相同的 identity，再被 latest-per-identity 挤掉。回放真实会话日志时，就有一条Updated instructions from: ~/.dsh/AGENTS.md，其正文同时含大写Instructions from:。src/config.ts：补文档——修订有独立 identity，所以"durable/agents-md"这个精确 takeover 不再覆盖修订；要收全家桶写"durable/*"，或再加上"durable/agents-md-update/*"。tests/issue2447-dsh-workspace-instructions.test.ts：翻转原第 138 行断言（原值undefined，旁边注明 "lowercase reconciliation variants stay consumable history"），并在旁边写清为什么翻转；补三种修订形态、多文件基线、正文引用大写标记的用例；新增端到端折叠场景。验证
session-b5ad5365-…6879 条事件、session-d979d0c3-…3277 条），把新正则直接跑在真实注入消息上：Additional instructions from: neighbor-help-frontend/AGENTS.md→durable/agents-md-update/neighbor-help-frontend/AGENTS.md；同时确认所有真实基线消息都不被修订正则误命中。上面那条"正文引用大写标记"的用例就是从这里来的，不是编的。正文引用大写标记那条用例靠INSTR_MSG.replace(...)注入，一旦 fixture 变动会静默不生效、断言照样通过。已加assert.notEqual(注入后, INSTR_MSG, "fixture injection must actually land")，注入没落地直接红。durable/agents-md；基线正文引用Instructions from:仍认作基线；修订正文引用它仍认作修订（顺序正确性）。compress让区间同时盖住两者 → 修订正文always run lint AND typecheck存活（它别处没有）；基线同时存活（独立 identity，没被修订挤下去）；区间里的填充消息确实被吃掉（证明折叠真发生了，不是保护把整个范围否决了）。src/durable-message-guards.ts重跑，issue2447的 5 个用例红 2 个（单测 + 新折叠场景）；恢复后 5/5 绿。确认这两个测试真钉住了行为，不是无论怎么实现都会通过的摆设。issue24475/5；issue2446/issue2481/issue276619/19；durable-tool-identity9/9；issue2419+protected-*系列 17/17（2 个 verif 用例无求解器环境自动 skip）；wire-contract-golden/config-surface-guard/marker-lane-coverage/golden-kernel-dist-guard全绿；npm run typecheck干净；unused-export / test-import-graph 两个 gate 干净。关联
Closes #4921
#2447（本条 durable 车道）· #2428（message-level 保护钩子）· #2419 / #2446 / #2481（同族 carrier）