Skip to content

fix: chain-detection content fallback exempts this instance's own sessions (#1086) - #1090

Merged
ranxianglei merged 4 commits into
masterfrom
2026-09-21_chain-content-self-exemption
Sep 22, 2026
Merged

ranxianglei merged 4 commits into
masterfrom
2026-09-21_chain-content-self-exemption

Conversation

@ranxianglei

@ranxianglei ranxianglei commented Sep 21, 2026 •

Copy link
Copy Markdown
Owner

Closes #1086, #1100

What happened

v0.1.133 (#1079) added a content fallback for bili→bili chain detection: when the body carries ACP artifacts (render tags or the ACP tool names), treat the request as an already-compressed chain payload and forward it verbatim — meant for middleboxes that strip x-bili-hop. Regression: those artifacts are exactly what bili itself injects/surfaces for its own managed clients (the opencode-native plugin registers acp_status/search_context in the client tool surface; clients re-send render tags verbatim). Single-instance setups therefore judged every turn a "chain", passed everything through, and the compression kernel never ran again → long sessions hit the upstream token limit with HTTP 400s. #1087 (v0.1.134) reverted the pair wholesale; this PR re-introduces the fallback correctly.

Design (4 layers)

  1. Structural detector (src/server/chain-artifacts.ts, new): allocation-free byte pre-filter, then structural verdict — tool declarations alone are NOT artifacts (that is the normal shape of a bili-managed client); historical tool calls of both names (all four wire shapes: OpenAI tool_calls[].function.name, Anthropic tool_use blocks, Responses function_call items, Gemini functionCall parts) or wire-escaped render tags ARE artifacts; prose never counts (JSON escaping defeats even the byte seed).
  2. Self-state exemption (src/session.ts:hasProcessedState): if THIS instance holds processed compression state for the session (memory first, then the persisted store — covers auto-update restarts: stats.requests>0 || nextBlockId>1 || blocks>0), the artifacts are self-produced ⇒ process normally. This is what v0.1.133 lacked: it judged before binding the session identity.
  3. Foreign verdict: artifacts + no local state ⇒ byte-identical passthrough, one warn per session (throttled, capped FIFO set), no session record created — same contract as the x-bili-hop path. Applies only when the store is enabled (see below); with persistence off the fallback is skipped entirely.
  4. Escape valve: chainContentDetection: false in the config file / BILI_CHAIN_CONTENT=0 env (env wins), documented in CONFIGURATION.md / CONFIGURATION.zh-CN.md. The x-bili-hop signal is unaffected by this switch.

The decision site is protocol-neutral (shared path before per-wire processing), so all four wire protocols get identical behavior.

Persistence-off instances (#1100)

The foreign verdict's "no local state ⇒ foreign" inference is only sound when persistence can prove ownership across a restart (loadSync restores disk state). With BILI_PERSIST=0 an instance cannot recover its sessions after a restart, so "no local state" is ambiguous with our OWN replayed history — and a decisive passthrough there would permanently disable compression for every own session after a restart (the #1086 symptom again, scoped to persistence-off + restart + replayed history). The fallback's byte pre-filter is therefore gated on getStore().enabled: when the store is disabled the whole content fallback is skipped and the request is processed normally. The x-bili-hop marker path is untouched and still catches real chains. (hasProcessedState already returned false in exactly this case, but that boolean cannot express "decline to make a verdict", so the decision belongs where the fallback is gated.)

Why not just keep the revert

Without the content fallback, a middlebox that strips x-bili-hop from a chained request silently re-enables double compression (the #292-class corruption the header exists to prevent). The protection stays; the misjudgment is fixed at its root (session identity + processed-state check + ownership-provability gate), not by removing the signal.

Residual corners (documented, bounded)

Evidence

  • npm run typecheck clean; npm test: 2171 pass / 0 fail / 2 skipped (e2e-gated); npm run build OK.
  • Regression suite tests/issue1086-chain-self-exemption.test.ts (9 tests): detector units across all four wire shapes + declaration-only (group E repro) + placeholder/prose negatives; T1 self-exemption under incident-shaped traffic (kernel rebuilt the body, zero [chain] warns); T2 fresh plugin-mode session processed; T3 foreign session byte-identical passthrough + single throttled warn + no session trace (now runs an enabled store over an empty temp dir, so "no local state" is a sound foreignness signal); T4 escape valve; T5 liveness guard — 60-turn openai-chat SSE loop against a relay that demands compress once context grows, asserting ≥1 kernel compress round-trip executed, bounded upstream bytes, and zero chain warns (the exact failure mode of the issue can no longer regress silently); T6 ([Bug] BILI_PERSIST=0 重启后自己的会话被永久判为 bili→bili 链(#1086 残留,#1090 随附) #1100) — BILI_PERSIST=0 + cleared memory (post-restart shape) replaying incident-shaped artifacts asserts the request is PROCESSED (kernel rebuilt the body, session created, stats advanced) NOT passed through verbatim, and that no chain warning fires (ownership unprovable ≠ confirmed-foreign). T6 fails on the pre-fix tree (passthrough) and passes post-fix.
  • Two adversarial review passes over the diff (correctness checklist + test-sufficiency checklist) completed before opening. The automated oracle route was unavailable in this environment (instant task failures), so both checklists were executed manually against the diff: decision placement vs prefix-affinity side effects (resolve() verified read-only apart from TTL expiry), forward() contract parity with the existing opencode v1: keep opencode-acp legacy sessions working after switching to bili #920 early-return bypass, first-request concurrency, per-request cost, config precedence vs sibling flags, style/compliance greps (hex escapes, no as any, version field untouched, no lockfile churn).

Auto-merge gate

Not auto-merge eligible: adds a config schema key and changes wire pass-through behavior (both compression modes). Human merge required.


修了什么问题:链检测的内容回退(#1079/v0.1.133)把 bili 自己注入的 ACP 产物(渲染标签、acp_status/search_context 工具面)误判为 bili→bili 链,导致单实例部署压缩内核永久跳过、长会话撞上游 1,048,576 上限报 400(#1086)。本 PR 在 #1087 整体回退的基础上正确重新引入该回退:结构性产物判定(声明≠产物)+ 本实例已处理状态豁免(内存/持久化存储)+ 外部载荷字节级透传且每会话仅告警一次 + 逃生阀 BILI_CHAIN_CONTENT=0。另修复深审发现的 F1(#1100):BILI_PERSIST=0 的实例重启后无法自证会话所有权,原逻辑会把回放历史的自有会话永久判为链而透传、压缩静默关闭——现将内容回退整体门控在 getStore().enabled 上,持久化关闭时直接走正常处理(x-bili-hop 路径不受影响)。回归测试含 60 轮长会话活性守卫(T5)与持久化关闭重启场景(T6)。可以合并。

@github-actions

github-actions Bot commented Sep 21, 2026 •

Copy link
Copy Markdown
Contributor

📦 Built Package Artifact

Branch: 2026-09-21_chain-content-self-exemption (60cfaa7)

Option A — Install from npm PR tag (recommended)

npm install -g billion-context@pr-1090

Each push to this PR publishes a new version under the pr-1090 npm tag.

Option B — Download artifact

  1. Download the artifact from the Actions run
  2. Extract the tarball and install:
tar xzf billion-context-pr1090.tgz
npm install -g package

Install instructions are refreshed on each push.

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷 Re-reviewed #1090 after pulling the branch. Verdict: mergeable — the code matches the #1086 spec and every check is green; I fixed one small docs defect directly on the branch.

Code review (against the #1086 repro) — correct and complete across all four layers:

  • Structural detector (src/server/chain-artifacts.ts): allocation-free byte pre-filter + JSON-aware render-tag regex (hex-escaped angle brackets, tolerates escaped attribute quotes) + history-only tool-name scan covering all four wire shapes (OpenAI chat / Responses / Anthropic / Gemini). Deliberately ignores the top-level tools declaration array ([Bug] 链检测把 bili 自己注入的 ACP 产物判成 bili→bili 链 —— 压缩内核永久跳过,长会话撞 1,048,576 上限 400(v0.1.133 回归) #1086 group E) and prose, so a fresh plugin-mode client that merely declares the tools is never misjudged.
  • Self-state exemption (src/session.ts hasProcessedState): memory first, persisted-store fallback (survives auto-update restart), read-only. Evidence = requests>0 ∨ nextBlockId>1 ∨ blocks>0.
  • Decision site (src/server.ts): the decision is deferred until session identity is resolved (anon-affinity vs conversation id) — exactly the gap v0.1.133 had. Foreign artifact request → byte-identical passthrough + one warn naming the escape valve; self → normal kernel processing. It returns before getSession, so foreign sessions leave no trace.
  • Escape valve (src/config.ts): chainContentDetection / BILI_CHAIN_CONTENT, default ON, env wins over file. x-bili-hop unaffected.

Verification:

Small defect fixed on the branch (review discipline: fix small problems directly, don't ping-pong): the two newly-added env-var table rows in CONFIGURATION.md and CONFIGURATION.zh-CN.md carried a stray leading space before | while every other row starts at column 0 → GitHub's GFM renderer would have drawn them outside the table. Removed the leading space on all four rows. Docs-only, zero code impact. New commit f3cdf65 pushed; CI re-running on the docs-only commit.

中文摘要:修复了 #1086 中链检测把本实例自身注入的 ACP 产物误判为 bili→bili 链、导致压缩内核整段停摆的问题——按已解析的会话身份区分"本实例自有会话"(正常压缩)与"外部接入请求"(原样透传+一次性警告),并保留 chainContentDetection/BILI_CHAIN_CONTENT 逃生阀;代码逐层核对无误,本地 typecheck+全量测试与全部 8 项 CI 均通过,另顺手修掉两个新增环境变量表格行因行首多余空格在 GitHub 上无法渲染的小问题(纯文档、零代码改动),可以合并。

Per repo rules the merge itself is human-only — please merge when ready: #1090

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷 Follow-up: CI on the new docs-only head f3cdf65 is now all 8 checks green — test ubuntu+windows × node 22/24, version-guard, build-artifact, real-codex zero-token e2e, and the CODEX_HOME overlay check. Nothing else pending on my side; ready for you to merge when convenient. 中文:新提交 f3cdf65(仅修文档表格行首空格)的 8 项 CI 已全部通过,无其他待办,可随时合并。

@ranxianglei ranxianglei left a comment

Copy link
Copy Markdown
Owner Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

审查结论:MERGE 级(人审合并;建议 F1/F2 各开 follow-up issue)

完整深审(代码逐层 + v0.1.133 旧检测器对拍 + 全部 7 个审查问题直答)后结论:可以合。要点:

已验证正确的核心

  • 4 层保真齐全,层(b)(artifact 结构性检测)比 PR body 承诺的 count-heuristic 更强;wire 纯读不改(chain-artifacts.ts 建 Set、链式路径 bodyBuffer 逐字节透传 prepared:null,T3 断言 sha 一致)。
  • TOCTOU 安全方向正确:false-"chain" 只导致 passthrough,永不丢数据;state 只由 processed 请求创建。
  • 免疫不变量在代码里成立:hop 标记只在 prepared !== null 时打(src/server.ts:3785→buildForwardTarget:3324),stats.requests 只在 processed 路径递增 → 内层 bili 转发体必带 hop,外层无 state → content fallback 正确触发。
  • 测试可证伪:T2/T1 对 a40fe47:src/server/chain-sig.ts:58-63(v0.1.133 全文 substring 匹配、无豁免)复现了原 bug。diff 卫生:7 文件 2 提交、无 version bump、hex 转义、zh/en 同步。

发现(按严重度)

  • F1 [Medium] BILI_PERSIST=0 重启即永久链式误判(src/session.ts:340-341 if (!store.enabled) return false):持久化关闭的实例重启后唯一自证证据被清空,tag 历史(plugin 客户端会回放)被永久 chain-flag,且 passthrough 不建 state、永不自愈——正是 #1086 的无界增长症状,只是范围缩小到 persistence-off 重启。默认 persistence-ON 大多数用户安全。建议 follow-up:store 禁用时内容 artifact 不作为决定性证据(跳过 content fallback,hop 保留)。
  • F2 [Medium] 磁盘分支零覆盖:4 个集成测试全部 _setStoreForTest({enabled:false})——store.loadSync 重启路径(注释自称覆盖 auto-update restart)、BILI_CHAIN_CONTENT env 解析、warn-set FIFO 淘汰都没测。建议 fast-follow:persist→flushAllSessions→_resetSessionsForTest→重放 artifact body→必须 process。
  • F3 [Medium 固有取舍] 迁移/换代理误报:bili-A 死亡后带 tag 历史迁到 bili-B 不可区分于剥头链;首请求就含用户粘贴的真形 tag(#1039 先例)同样误判 → 永久 passthrough、压缩静默关闭。缓解已到位(严格 regex + 每会话一次 warn 带 knob 提示 + 逃生阀),README troubleshooting 值得加一行。
  • F4 [Low 已知边界] 首请求剥 hop 且无 artifact 的真链会被外层 processed → 双压缩;0.1.132 hop-only 同洞,非回归。
  • F5 [Low] 请求路径同步 IO:仅 byte-seed 命中+内存 miss 后触发;外会话两次 ENOENT stat。可接受。
  • F6 [Nit] loadSync 在 #408 负 token 钳制时 scheduleSave——"只读"探针带写副作用;沿袭既有行为,无害。

按 AGENTS §7.4 本 PR 改 wire 行为(chain 判定)须人审合并——与计划一致。建议合并后把 F1、F2 各开 issue。

@ranxianglei

Copy link
Copy Markdown
Owner Author

@ework — please re-review the current head f3cdf65 in full. Additional context since your last pass: an independent deep review is at #1090 (review) (verdict: merge-grade, F1–F6 findings; F1/F2 filed as #1100 / #1101). Re-verify against those findings and the follow-up issues before human merge.

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷 Full re-review of head f3cdf65 against the independent deep review (review-5268457270, F1–F6). I re-read the cited code myself rather than trusting the summary. Verdict: MERGE-grade — unchanged. No finding blocks f3cdf65; F1/F2 are real and already have active follow-ups (#1100/#1101), F3–F6 are inherent/low/nit/pre-existing.

Base & gates

  • Branch 2026-09-21_chain-content-self-exemption @ f3cdf65 = 2 commits ahead of current master, 0 behind → no rebase needed.
  • CI on f3cdf65: all 8 checks green (test ubuntu/windows × node 22/24, version-guard, build-artifact, real-codex zero-token e2e, CODEX_HOME overlay). Prior local pass: typecheck ✅, regression 8/8, full suite 2170/0/2.
  • Diff hygiene held: 7 files, no version bump, hex-escaped <acp>, zh/en synced.

Independent verification of each finding

Core invariants re-confirmed by reading the code: identity resolved before the verdict (sessionId at src/server.ts:1421, consumed at :1431); foreign → byte-identical passthrough returning before getSession so no state/trace is created (:1439-1440); self → normal processing (:1442-1443); immunity invariant — hop marker stamped only on actually-processed forwards (prepared !== null ? instanceId : undefined at :3796 → buildForwardTarget :3324). Wire stays pure-read; the detector ignores top-level tool declarations and prose.

Recommendation
f3cdf65 is independently correct and green, so it is mergeable as-is. Because #1101 lands its restart-path tests on this same branch, the cleaner path is: let that commit land first (it re-triggers CI), confirm green on the new head, then merge once so the feature ships together with the coverage that verifies the very restart path the code comment claims. #1100's fix should ship as its own reviewed PR. Your call either way — nothing here is a correctness blocker for f3cdf65.

中文摘要:逐条对照独立深审的 F1–F6 并亲自读取被引用的代码复核——F1(BILI_PERSIST=0 重启后自证证据丢失、回放历史被永久误判为链而压缩静默关闭)与 F2(磁盘 loadSync 重启路径 / BILI_CHAIN_CONTENT 解析 / warn 队列 FIFO 淘汰零测试覆盖)均属实,且已分别由 #1100 / #1101 跟进(#1101 的测试会直接加到本分支),F3–F6 属固有取舍 / 低危 / nit / 既有边界,均不构成阻塞;核心不变量(会话身份先于判定、外部请求无状态透传、hop 标记仅在实际处理时打)在代码层面成立,f3cdf65 可合并;建议等 #1101 的测试提交落到同一分支且 CI 转绿后一次性合并,#1100 的行为修复走独立的评审 PR,合并动作仍由人工执行。

Per repo rules the merge itself is human-only: #1090

ework-agent added 4 commits September 22, 2026 18:35
…sions (#1086)

v0.1.133's content fallback judged bili's own injected ACP artifacts (render tags / acp_status+search_context tool surface) as bili->bili chain evidence and passed every turn through verbatim, so the compression kernel never ran again on single-instance setups (#1086). Re-introduces the fallback correctly after #1087's revert:

- structural detector: declarations alone are not artifacts; historical tool calls (both names, all four wire shapes) or wire-escaped render tags are; prose never counts
- self-state exemption: processed local state for the session (memory or persisted store) means the artifacts are self-produced, process normally
- foreign artifacts without local state: byte-identical passthrough, one warn per session, no session record created
- escape valve: chainContentDetection:false / BILI_CHAIN_CONTENT=0
…1100)

The chain-detection content fallback (#1086) decides "no local processed
state => foreign bili->bili chain => passthrough". That inference is only
sound when persistence can prove ownership across a restart: with
BILI_PERSIST=0 an instance cannot recover its session state after a
restart, so "no local state" is ambiguous with our OWN replayed session.
Treating it as decisive made a persistence-off instance permanently
pass-through every own session after a restart — compression silently off
and unbounded growth, i.e. the #1086 symptom again, scoped to
persistence-off + restart + replayed history.

Gate the fallback's byte pre-filter on getStore().enabled: when the store
is disabled the whole content fallback is skipped and the request is
processed normally. The x-bili-hop marker path is untouched and still
catches real chains. hasProcessedState already returned false in exactly
this case, but that boolean cannot express "decline to make a verdict", so
the decision belongs at the point where the fallback is gated.

Tests: T3 (foreign => passthrough) now runs an ENABLED store over an empty
temp dir so "no local state" is a sound foreignness signal; new T6 asserts
the BILI_PERSIST=0 + restart replay is processed, not passed through.

Closes #1100
@ranxianglei
ranxianglei force-pushed the 2026-09-21_chain-content-self-exemption branch from ec384a2 to 60cfaa7 Compare September 22, 2026 10:37
@ranxianglei

Copy link
Copy Markdown
Owner Author

Rebased onto current master (945a161, includes #1119/#1120 image-aware estimator + #1137 anon-affinity fix). The branch had grown 2 commits since first review (#1100 persistence gate, #1101 tests) — all 4 reviewed on the rebased head.

Rebase notes:

Gates on the rebased head: typecheck ✅ · full suite 2230 pass / 0 fail / 2 skipped (E2E gates) ✅ · build ✅.

Closes #1086, closes #1100, closes #1101.

@ranxianglei

Copy link
Copy Markdown
Owner Author

🤖 Powered by ework · qwen3.8-27b

[bot] 🏷 Re-reviewed the rebased head 60cfaa7 in full. Verdict: MERGE-READY — the rebase is clean, both folded-in commits are correct, and every gate is green. Closes #1086 + #1100 + #1101.

Rebase hygiene (§7.2's #1 rework driver — checked first)

  • Base 945a161 == current origin/master → branch is exactly 4 ahead / 0 behind; no drift, no second rebase needed. mergeable: True (no conflict).
  • The one flagged conflict (src/server.ts imports) resolved correctly by union: line 113 keeps master's stream-io import incl. awaitDrain (still used at :4601), line 114 adds this PR's chain-artifacts import — no duplicate or dropped identifier. Typecheck confirms it compiles.
  • Diff vs base: 7 files, 838 ins / 1 del (near-purely additive); no package.json/version bump (master stays at 0.1.137), hex-escaped <acp>, zh/en synced.

Folded commit #1100 (245a4bd) — the one new behavior change, reviewed against code

  • Gate sits at the seed computation: artifactSeed = … && opts.chainContentDetection !== false && getStore().enabled && artifactSeedHit(bodyBuffer). With BILI_PERSIST=0 the whole content fallback is skipped → request processed normally; the x-bili-hop marker path above it is untouched and still catches real chains.
  • Correct call, well-reasoned: "no local state ⇒ foreign" is only sound when persistence can prove ownership across a restart; persist-off can't, so "no state" is ambiguous with our own replayed session. Placing the gate here (rather than overloading hasProcessedState's boolean) is the right layer — that boolean can't express "decline to verdict." Fixes F1 with zero default-config regression (persist-on path unchanged).
  • Tradeoff I explicitly checked: persist-off + a genuine stripped-hop chain will now process (a rare mild double-compress) instead of passthrough. In persist-off there is no signal to distinguish own-replay from foreign-stripped-chain, so some default is forced; defaulting to "process" trades a rare mild double-compress for eliminating total compression-death — the lesser evil, and it's documented in the commit. Real chains usually keep their hop marker, narrowing it further. Acceptable.
  • Folding [Bug] BILI_PERSIST=0 重启后自己的会话被永久判为 bili→bili 链(#1086 残留,#1090 随附) #1100 into this PR (vs my earlier "ship it separately" suggestion) is actually better: landing the fallback without its persist-off guard would reopen F1 for that subset in a release window; co-delivering closes the edge atomically.

Folded commit #1101 (60cfaa7) — closes all three F2 gaps

  • Its src/server.ts touch is pure test-enablement: exports WARNED_CHAIN_SESSION_CAP and adds _chainWarnSetForTest() (read-only getter). No behavior change hidden under a test: label.
  • T7 exercises the previously-untested disk branch of hasProcessedState (src/session.ts:342): warmup → persist → simulated restart (fresh store over the same dir) → replay, asserting processed-not-passthrough and stats.requests 1→2, which proves reload-from-disk rather than a freshly created session. That validates exactly the auto-update-restart guarantee the comment claims.
  • T8 asserts FIFO evicts exactly the oldest entry past the cap (new session tracked, pad-0 gone, pad-1 kept).
  • T9 covers env-parse precedence: default ON, BILI_CHAIN_CONTENT=0 disables, env beats file.
  • Also confirmed T3 was correctly repurposed to an enabled store over an empty temp dir (a genuinely-foreign case), since post-[Bug] BILI_PERSIST=0 重启后自己的会话被永久判为 bili→bili 链(#1086 残留,#1090 随附) #1100 a disabled store no longer exercises the foreign-passthrough path.

Gates on 60cfaa7

  • CI: all 8 checks green (test ubuntu/windows × node 22/24, version-guard, build-artifact, real-codex zero-token e2e, CODEX_HOME overlay). Any mergeable_state: blocked seen right after the push was just CI still running — it's green now, and with mergeable: True the PR is ready.
  • Local independent run: typecheck ✅ · regression file 12/12 (T1–T5 + T6/T7/T8/T9 + detector units) ✅ · build ✅.

中文摘要:对 rebase 后的 head 60cfaa7 做了完整复审——rebase 干净(base 即当前 master、无冲突、union 合并的 import 无重复或丢失标识符)、diff 近纯新增(7 文件 838+/1-)且未动 version;新并入的 #1100 在 content-fallback 的字节预筛上加了 getStore().enabled 门(BILI_PERSIST=0 时跳过整个回退并正常处理,hop 标记路径不受影响),正确修复 F1 且默认配置零回归(唯一代价是 persist-off + 真实 strip-hop 链会偶发一次轻微双重压缩,属更轻的取舍且已在提交中记录);#1101 的 T7/T8/T9 补齐了磁盘重启分支 / warn 队列 FIFO 淘汰 / env 优先级这三处此前零覆盖的测试,其对 server.ts 的改动仅为测试导出、非行为变更;CI 8 项全绿、本地 typecheck / 回归 12/12 / build 均通过,可合并,合并后将同时关闭 #1086 / #1100 / #1101,合并动作仍由人工执行。

Per repo rules the merge itself is human-only: #1090

@ranxianglei
ranxianglei merged commit ecc7bb1 into master Sep 22, 2026
8 checks passed
ranxianglei pushed a commit that referenced this pull request Sep 23, 2026
…le audit log (#1082)

Rebased onto master v0.1.139 (b2f28c6): adopt createStorageCodec({key})
in tests (codec rename from the #1083 era), union #1090 hasProcessedState
beside dropSessionForGc, keep master's updated BILI_ENCRYPTION_KEY doc
row (drop pre-#1083 wording). Squashed of the five prior commits.
@ranxianglei
ranxianglei deleted the 2026-09-21_chain-content-self-exemption branch October 15, 2026 05:32
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

1 participant