惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

F
Full Disclosure
Recorded Future
Recorded Future
T
Tenable Blog
S
Securelist
C
CERT Recently Published Vulnerability Notes
T
Threatpost
S
Schneier on Security
A
Arctic Wolf
The Hacker News
The Hacker News
C
CXSECURITY Database RSS Feed - CXSecurity.com
Know Your Adversary
Know Your Adversary
P
Privacy International News Feed
Threat Intelligence Blog | Flashpoint
Threat Intelligence Blog | Flashpoint
The Register - Security
The Register - Security
Cisco Talos Blog
Cisco Talos Blog
AWS News Blog
AWS News Blog
K
Kaspersky official blog
T
True Tiger Recordings
T
Threat Research - Cisco Blogs
V
Vulnerabilities – Threatpost
P
Palo Alto Networks Blog
T
The Exploit Database - CXSecurity.com
小众软件
小众软件
B
Blog
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
Microsoft Azure Blog
Microsoft Azure Blog
Cyberwarzone
Cyberwarzone
C
Cybersecurity and Infrastructure Security Agency CISA
T
Tor Project blog
Spread Privacy
Spread Privacy
Malwarebytes
Malwarebytes
P
Proofpoint News Feed
F
Fox-IT International blog
F
Fortinet All Blogs
P
Privacy & Cybersecurity Law Blog
G
GRAHAM CLULEY
量子位
Latest news
Latest news
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
博客园 - 叶小钗
Project Zero
Project Zero
T
Tailwind CSS Blog
N
Netflix TechBlog - Medium
Martin Fowler
Martin Fowler
IntelliJ IDEA : IntelliJ IDEA – the Leading IDE for Professional Development in Java and Kotlin | The JetBrains Blog
IntelliJ IDEA : IntelliJ IDEA – the Leading IDE for Professional Development in Java and Kotlin | The JetBrains Blog
I
Intezer
博客园_首页
腾讯CDC
H
Hackread – Cybersecurity News, Data Breaches, AI and More
D
Darknet – Hacking Tools, Hacker News & Cyber Security

Recent Commits to openclaw:main

test: fix current suite drift · openclaw/openclaw@80655fe fix(lock): require owner identity proof before stale removal · openclaw/openclaw@daa7b1d fix(deps): pin shrinkwrap patch drift to pnpm lock · openclaw/openclaw@d8a14e7 revert: 60bec8c duplicate tool display guard · openclaw/openclaw@e09f89d fix(e2e): bound docker package preparation · openclaw/openclaw@38edae7 fix(cli): add Windows stack-size respawn (#87031) · openclaw/openclaw@5e8f498 fix(agents): guard duplicate tool display metadata (#87025) · openclaw/openclaw@60bec8c ci(release): port 2026.5.25 release gate fixes · openclaw/openclaw@f7e2d9b chore: update tool display snapshot · openclaw/openclaw@ad71c42 fix(web-search): keep runtime legacy merge out of validation (#86818) · openclaw/openclaw@4a85cd7 fix(cli): default logs to local timestamps (#85387) · openclaw/openclaw@3127808 fix(agents): dedupe transcripts tool display config · openclaw/openclaw@8788ae1 fix(updater): exclude prerelease tags from stable git channel (#86559) · openclaw/openclaw@e070519 fix(agents): memoize session lock owner args · openclaw/openclaw@c430fcd fix: dedupe transcripts tool display metadata · openclaw/openclaw@0f49bbb fix(cli): validate timeout and banner TTY state · openclaw/openclaw@abb85cc fix(codex): project newer history on app-server resume (#86677) fix(codex): keep turn timeouts inside Codex (#86476) fix(auto-reply): stage sandboxed workspace media · openclaw/openclaw@f22c3a5 fix(e2e): support plain telegram install timeouts fix(mac): harden restart and dSYM packaging · openclaw/openclaw@639e7ff fix(exec): avoid default approval store writes (#86964) · openclaw/openclaw@4d65936 fix(agents): restore current guard checks (#86934) · openclaw/openclaw@9b1b6d0 docs(changelog): prepare 2026.5.26 notes · openclaw/openclaw@983b338 fix(commands): keep slash handling off reply startup · openclaw/openclaw@c2b56de fix(reply): defer context compaction safely · openclaw/openclaw@ed3ae0d fix(telegram): refine typing and progress drafts · openclaw/openclaw@0afccc6 fix(codex): gate profiler timing and startup setup · openclaw/openclaw@21c25bb fix(agents): avoid runtime model hydration on hot paths · openclaw/openclaw@7951cc0 fix(reply): reduce visible reply delivery latency · openclaw/openclaw@699c047 docs(changelog): note reply latency fixes · openclaw/openclaw@29a1dc2 fix(e2e): support plain timeout wrappers fix(channels): preserve direct native progress callbacks · openclaw/openclaw@e750041 fix: tighten parser edge cases (#86999) · openclaw/openclaw@174cd49 fix(e2e): clean stale docker lane containers · openclaw/openclaw@3968288 fix(e2e): bound docker lifecycle hangs · openclaw/openclaw@71cb607 fix(gateway): bound live agent model probes fix(e2e): bound plugin binding docker smoke · openclaw/openclaw@b36fa1d fix(e2e): preserve docker run failure status feat(discord): bucket large model picker menus fix(telegram): derive DM topics from bot capability · openclaw/openclaw@aa117ec fix: improve discord voice playback and wake replies fix(e2e): kill timed kitchen rpc command groups · openclaw/openclaw@23aeb58 ci: use supported codex mini live target · openclaw/openclaw@b56ddcc fix(ci): kill wedged bun smoke commands fix(e2e): bound corrupt plugin update runs · openclaw/openclaw@2b63eb2 ci: require codex profiles for live probes · openclaw/openclaw@6930538 docs: clarify inline comment guidance docs: update changelog for landed sweep fixes · openclaw/openclaw@8c575bd fix(agents): disclose scoped session list results (#86944) · openclaw/openclaw@598aad4 fix(telegram): treat ENETDOWN as transient network failure (#86762) · openclaw/openclaw@1fd8de8 fix(mac): harden package script safety · openclaw/openclaw@564e0bb fix(ci): kill wedged checkout fetches · openclaw/openclaw@c867ecb ci: restore codex replay live probe contract · openclaw/openclaw@9fd8158 fix(codex): preserve oversized native thread reuse · openclaw/openclaw@7a14741 fix(scripts): detect timed changed gates · openclaw/openclaw@a5eee8f ci: stop waiting for nonexistent capability restart wake · openclaw/openclaw@3c6fd49 fix(e2e): route plugin update through timeout helper · openclaw/openclaw@e8f584e fix: respect root options in startup guards (#86927) · openclaw/openclaw@7e6837b fix: tighten CLI utility failure handling (#86918) · openclaw/openclaw@0ec2928 fix: preserve config and hook contracts (#86911) fix: tighten small runtime parsing guards (#86909) fix(ci): kill timed tui pty test runs · openclaw/openclaw@081e295 fix(logging): preserve env placeholders during redaction · openclaw/openclaw@6c18c21 fix(memory-core): avoid per-file watcher FD fan-out for memory direct… fix: use current config sdk contract in feishu doctor · openclaw/openclaw@5535eef Validate wide-area DNS zone domains [AI] (#84136) · openclaw/openclaw@84b9704 ci: stabilize release live QA gates · openclaw/openclaw@27359ec ci: harden live release gates · openclaw/openclaw@cf21c8a ci: normalize Windows toolcache Node paths perf: cache read-only channel resolution chore: remove stale codex test conversion · openclaw/openclaw@fdb7848 fix(test): explain missing vitest dependency · openclaw/openclaw@373b3bf fix(e2e): kill timed docker scenario runners · openclaw/openclaw@d5bf325 fix: add transcripts tool display metadata · openclaw/openclaw@645cbf6 docs: update changelog for landed fixes · openclaw/openclaw@12b81d8 fix(agents): route btw through embedded stream resolver (#86312) fix(telegram): treat targeted bot commands as mentions (#86553) · openclaw/openclaw@c7821bd fix(e2e): route doctor switch commands through timeout helper · openclaw/openclaw@9ced76a fix: ignore other codex thread completions · openclaw/openclaw@ead847f fix(feishu): repair stale channel state · openclaw/openclaw@b7c461a fix: remove stale image provider assertions · openclaw/openclaw@0973a7e fix: accept trailing fuzzy voice wake questions · openclaw/openclaw@d001d35 fix(podman): bound setup image builds · openclaw/openclaw@d6fcb56 fix(podman): kill timed container launches · openclaw/openclaw@6118f3f fix(scripts): preserve native pnpm exec paths · openclaw/openclaw@fb853de fix(codex): bridge cli api-key auth into app-server fix: update Discord voice to libopus-wasm 0.1.0 · openclaw/openclaw@0f605ee fix(e2e): kill timed docker helper commands · openclaw/openclaw@e89afa6 fix(e2e): kill timed live docker runs · openclaw/openclaw@dc0d4c2 fix(ci): kill timed website installer docker steps · openclaw/openclaw@d54c906 fix(scripts): trim macOS node bootstrap · openclaw/openclaw@4ff5a61 fix(ci): kill timed install smoke docker steps · openclaw/openclaw@cf6f9ad fix(setup): kill timed image pulls when supported fix(test): default Vitest stall watchdog fix(ci): bound crabbox hydrate downloads · openclaw/openclaw@d1c8f09 fix(control-ui): guard stale overview usage refresh fix(qa): require genai otel model spans (#86920) refactor: move transcripts into core fix(message-tool): hydrate structured reply attachments
fix(agents): preserve sessions_spawn transcript payloads (#82203) · openclaw/openclaw@ef86d8c
martingarram · 2026-05-27 · via Recent Commits to openclaw:main

@@ -2321,7 +2321,109 @@ describe("wrapStreamFnSanitizeMalformedToolCalls", () => {

23212321

]);

23222322

});

232323232324-

it("drops signed thinking turns when replay would expose inline sessions_spawn attachments", async () => {

2324+

it("keeps signed thinking turns that reuse a mutable earlier tool id", async () => {

2325+

const messages = [

2326+

{

2327+

role: "assistant",

2328+

content: [{ type: "toolCall", id: "call_1", name: "read", arguments: {} }],

2329+

},

2330+

{

2331+

role: "toolResult",

2332+

toolCallId: "call_1",

2333+

content: [{ type: "text", text: "mutable result" }],

2334+

},

2335+

{

2336+

role: "assistant",

2337+

content: [

2338+

{ type: "thinking", thinking: "internal", thinkingSignature: "sig_1" },

2339+

{ type: "toolUse", id: "call_1", name: "read", input: {} },

2340+

],

2341+

},

2342+

{

2343+

role: "toolResult",

2344+

toolCallId: "call_1",

2345+

content: [{ type: "text", text: "signed result" }],

2346+

},

2347+

{

2348+

role: "user",

2349+

content: [{ type: "text", text: "retry" }],

2350+

},

2351+

];

2352+

const baseFn = vi.fn((_model, _context) =>

2353+

createFakeStream({ events: [], resultMessage: { role: "assistant", content: [] } }),

2354+

);

2355+2356+

const wrapped = wrapStreamFnSanitizeMalformedToolCalls(baseFn as never, new Set(["read"]), {

2357+

validateAnthropicTurns: true,

2358+

preserveSignatures: true,

2359+

dropThinkingBlocks: false,

2360+

} as never);

2361+

const stream = wrapped(

2362+

{ api: "anthropic-messages" } as never,

2363+

{ messages } as never,

2364+

{} as never,

2365+

) as FakeWrappedStream | Promise<FakeWrappedStream>;

2366+

await Promise.resolve(stream);

2367+2368+

expect(baseFn).toHaveBeenCalledTimes(1);

2369+

const seenContext = firstBaseContext(baseFn);

2370+

expect(seenContext.messages).toBe(messages);

2371+

});

2372+2373+

it("drops signed thinking reused ids when their real result is displaced", async () => {

2374+

const firstAssistant = {

2375+

role: "assistant",

2376+

content: [{ type: "toolCall", id: "call_1", name: "read", arguments: {} }],

2377+

};

2378+

const firstResult = {

2379+

role: "toolResult",

2380+

toolCallId: "call_1",

2381+

toolName: "read",

2382+

content: [{ type: "text", text: "mutable result" }],

2383+

};

2384+

const userMessage = {

2385+

role: "user",

2386+

content: [{ type: "text", text: "retry" }],

2387+

};

2388+

const messages = [

2389+

firstAssistant,

2390+

firstResult,

2391+

{

2392+

role: "assistant",

2393+

content: [

2394+

{ type: "thinking", thinking: "internal", thinkingSignature: "sig_1" },

2395+

{ type: "toolUse", id: "call_1", name: "read", input: {} },

2396+

],

2397+

},

2398+

userMessage,

2399+

{

2400+

role: "toolResult",

2401+

toolCallId: "call_1",

2402+

content: [{ type: "text", text: "signed result" }],

2403+

},

2404+

];

2405+

const baseFn = vi.fn((_model, _context) =>

2406+

createFakeStream({ events: [], resultMessage: { role: "assistant", content: [] } }),

2407+

);

2408+2409+

const wrapped = wrapStreamFnSanitizeMalformedToolCalls(baseFn as never, new Set(["read"]), {

2410+

validateAnthropicTurns: true,

2411+

preserveSignatures: true,

2412+

dropThinkingBlocks: false,

2413+

} as never);

2414+

const stream = wrapped(

2415+

{ api: "anthropic-messages" } as never,

2416+

{ messages } as never,

2417+

{} as never,

2418+

) as FakeWrappedStream | Promise<FakeWrappedStream>;

2419+

await Promise.resolve(stream);

2420+2421+

expect(baseFn).toHaveBeenCalledTimes(1);

2422+

const seenContext = firstBaseContext(baseFn);

2423+

expect(seenContext.messages).toEqual([firstAssistant, firstResult, userMessage]);

2424+

});

2425+2426+

it("drops signed thinking turns with inline sessions_spawn attachments when the result is missing", async () => {

23252427

const attachmentContent = "SIGNED_THINKING_INLINE_ATTACHMENT";

23262428

const messages = [

23272429

{

@@ -2374,7 +2476,7 @@ describe("wrapStreamFnSanitizeMalformedToolCalls", () => {

23742476

]);

23752477

});

237624782377-

it("drops signed thinking turns when replay would expose non-content attachment payload fields", async () => {

2479+

it("drops signed thinking turns with non-content attachment payload fields when the result is missing", async () => {

23782480

const attachmentContent = "SIGNED_THINKING_NESTED_ATTACHMENT";

23792481

const messages = [

23802482

{

@@ -2433,6 +2535,60 @@ describe("wrapStreamFnSanitizeMalformedToolCalls", () => {

24332535

]);

24342536

});

243525372538+

it("keeps signed thinking turns with sessions_spawn attachments when the tool result is present", async () => {

2539+

const attachmentContent = "SIGNED_THINKING_PAIRED_ATTACHMENT";

2540+

const messages = [

2541+

{

2542+

role: "assistant",

2543+

content: [

2544+

{ type: "thinking", thinking: "internal", thinkingSignature: "sig_1" },

2545+

{

2546+

type: "toolUse",

2547+

id: "call_1",

2548+

name: "sessions_spawn",

2549+

input: {

2550+

task: "inspect attachment",

2551+

attachments: [{ name: "snapshot.txt", content: attachmentContent }],

2552+

},

2553+

},

2554+

],

2555+

},

2556+

{

2557+

role: "toolResult",

2558+

toolCallId: "call_1",

2559+

toolName: "sessions_spawn",

2560+

content: [{ type: "text", text: "done" }],

2561+

},

2562+

{

2563+

role: "user",

2564+

content: [{ type: "text", text: "retry" }],

2565+

},

2566+

];

2567+

const baseFn = vi.fn((_model, _context) =>

2568+

createFakeStream({ events: [], resultMessage: { role: "assistant", content: [] } }),

2569+

);

2570+2571+

const wrapped = wrapStreamFnSanitizeMalformedToolCalls(

2572+

baseFn as never,

2573+

new Set(["sessions_spawn"]),

2574+

{

2575+

validateAnthropicTurns: true,

2576+

preserveSignatures: true,

2577+

dropThinkingBlocks: false,

2578+

} as never,

2579+

);

2580+

const stream = wrapped(

2581+

{ api: "anthropic-messages" } as never,

2582+

{ messages } as never,

2583+

{} as never,

2584+

) as FakeWrappedStream | Promise<FakeWrappedStream>;

2585+

await Promise.resolve(stream);

2586+2587+

expect(baseFn).toHaveBeenCalledTimes(1);

2588+

const seenContext = firstBaseContext(baseFn);

2589+

expect(seenContext.messages).toBe(messages);

2590+

});

2591+24362592

it("keeps mutable thinking turns outside anthropic replay-only preservation", async () => {

24372593

const messages = [

24382594

{

@@ -3044,10 +3200,6 @@ describe("wrapStreamFnSanitizeMalformedToolCalls", () => {

30443200

messages: Array<{ role?: string; content?: unknown[] }>;

30453201

};

30463202

expect(seenContext.messages).toEqual([

3047-

{

3048-

role: "assistant",

3049-

content: [{ type: "text", text: "[tool calls omitted]" }],

3050-

},

30513203

{

30523204

role: "user",

30533205

content: [{ type: "text", text: "retry" }],