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

fix(e2e): bound Telegram credential setup · openclaw/openclaw@9b28603 perf(gateway): borrow agent session lookup safely · openclaw/openclaw@3b9fa16 fix(crabbox): prefer Azure for Windows targets (#87186) · openclaw/openclaw@6afe3e8 test(e2e): isolate shell helper env · openclaw/openclaw@b28f9e0 fix(e2e): bound MCP channel connect · openclaw/openclaw@aa40174 perf(gateway): borrow read-only session metadata · openclaw/openclaw@8c8162f fix(gateway): persist model auth profile suffixes · openclaw/openclaw@152f68d fix(googlechat): suppress thread sends in DMs · openclaw/openclaw@da822dd Fix embedded session file ownership race (#87159) · openclaw/openclaw@3349fe2 ci(ui): refresh raw copy baseline · openclaw/openclaw@ebe09be fix(pi-embedded): route Codex OAuth compaction through OpenAI-Codex · openclaw/openclaw@1710dac fix(commands): enforce /allowlist configWrites origin policy · openclaw/openclaw@ce64d74 fix(plugin-sdk): stop exporting vitest test helpers (#87120) · openclaw/openclaw@780bc79 perf(gateway): borrow session reads on turn hot paths · openclaw/openclaw@9c2a6a8 fix(reply): keep visible turn admission unbounded (#87044) · openclaw/openclaw@455d5e8 fix(e2e): bound Telegram proof Bot API calls · openclaw/openclaw@4b40197 test(core): isolate provider and approval tests · openclaw/openclaw@e7214ef fix(agents): preserve sessions fallback errors · openclaw/openclaw@7986917 fix(prompt): route untrusted group prompts outside system prompt [AI]… · openclaw/openclaw@2c88547 feat: expose plugin approval action metadata · openclaw/openclaw@0c867ee docs(changelog): refresh 2026.5.26 notes · openclaw/openclaw@e74a039 fix: bind plugin command llm auth to host agent · openclaw/openclaw@b9f6abf fix(e2e): bound kitchen sink RPC probes · openclaw/openclaw@96bd939 ci: support native Windows Crabbox daemon hydration (#87175) · openclaw/openclaw@95c8fc9 fix: handle sessions_send active fallback failures (#86638) · openclaw/openclaw@81e7e8e fix(gateway): enable default auth rate limiting (#87148) · openclaw/openclaw@ae972fe test(media): isolate generation provider registry mocks (#87173) · openclaw/openclaw@9772cf2 perf(gateway): skip lifecycle session cache clone · openclaw/openclaw@0762aca fix(e2e): backstop Parallels update jobs · openclaw/openclaw@fdf58c1 fix(daemon): keep Windows Scheduled Task running on battery power (#5… · openclaw/openclaw@d92a333 fix(daemon): avoid workgroup schtasks user prompts · openclaw/openclaw@f348284 perf(gateway): avoid cloning live switch store reads · openclaw/openclaw@b75f70b fix(crabbox): use host-visible local work roots · openclaw/openclaw@586a6ce docs(changelog): refresh 2026.5.26 notes · openclaw/openclaw@15c0dfa fix(exec): hide unavailable durable approval actions (#86359) · openclaw/openclaw@42f0822 fix(reply): derive explicit control command turns · openclaw/openclaw@2899560 fix(e2e): check onboarding systemd noise · openclaw/openclaw@44c1cc8 test(agents): cover cold default model alias resolution · openclaw/openclaw@2e3b4b5 fix: keep plugin runtime fallback on startup root · openclaw/openclaw@2b9be22 fix: expand startup argv runtime fallback hints · openclaw/openclaw@e71b6f7 test: cover plugin runtime diagnostic context · openclaw/openclaw@78b2aee Fix runtime fallback startup argv default · openclaw/openclaw@66a8262 fix: prefer trusted argv runtime fallback roots · openclaw/openclaw@5371b96 Fix plugin runtime module resolution diagnostics · openclaw/openclaw@41fa603 fix(ui): show config open failure feedback (#87108) · openclaw/openclaw@8246e91 fix(e2e): bound Telegram RTT bot API calls · openclaw/openclaw@5981822 fix(install): bound finalization probes (#86997) · openclaw/openclaw@bf1a5c3 fix(memory): reject prompt-like memory stores (#87142) · openclaw/openclaw@119d235 fix(e2e): bound release user journey fixture probes · openclaw/openclaw@6b68d05 fix(e2e): bound bundled runtime HTTP probes · openclaw/openclaw@d886816 perf(gateway): skip duplicate turn session touch · openclaw/openclaw@8fa4fad test: avoid repeated module reloads in unit tests · openclaw/openclaw@1c8a112 fix(media): keep explicit workspace roots scoped · openclaw/openclaw@608fa52 fix(e2e): bound bundled runtime smoke commands · openclaw/openclaw@fca77dc test: route more command tests through light suite · openclaw/openclaw@bbfcdea fix(scripts): short-circuit helper help · openclaw/openclaw@4b23b36 test: harden docker smoke portability · openclaw/openclaw@10056c9 fix(agents): recover failed subagent lifecycle completions · openclaw/openclaw@4980c32 fix(e2e): hard kill timed out host commands · openclaw/openclaw@dd44a47 test: move lightweight command tests to light suite · openclaw/openclaw@2831d69 fix(scripts): handle helper cli help · openclaw/openclaw@2cc6871 fix(gateway): bound loopback preflight calls · openclaw/openclaw@6d5c15a fix(hooks): enforce default hook agent allowlist · openclaw/openclaw@e72621e fix(e2e): handle docker helper cli help · openclaw/openclaw@2814ab6 fix(codex): avoid native compaction on budget triggers (#86772) · openclaw/openclaw@eb8f9b4 test: speed up plugin test fixtures · openclaw/openclaw@05ff771 fix(e2e): bound openai chat tools client · openclaw/openclaw@de94217 fix(e2e): bound upgrade survivor probes · openclaw/openclaw@981ae13 fix(doctor): keep hooks model checks read-only (#86101) · openclaw/openclaw@371c4d6 fix(installer): tighten nonroot smoke node preflight · openclaw/openclaw@340f480 fix(e2e): bound HTTP readiness probes · openclaw/openclaw@d58f864 fix(daemon): keep node tasks off gateway listener cleanup · openclaw/openclaw@a4e0b6e perf(gateway): reuse prepared auth stores · openclaw/openclaw@d2711c9 test: speed up codex app server run attempts · openclaw/openclaw@1ce3637 build(codex): update Codex CLI to 0.134.0 · openclaw/openclaw@231a812 docs(skills): omit advisory ids from changelog notes · openclaw/openclaw@989a369 test: speed up test project routing · openclaw/openclaw@140892c Fix stale approval prompts in Control UI (#86270) · openclaw/openclaw@5297eeb fix(installer): reject stale cli node runtimes · openclaw/openclaw@49d605e test: harden e2e harness isolation · openclaw/openclaw@acbb06e fix: keep approval runtime token local-only · openclaw/openclaw@96c5766 perf(gateway): defer skipped-channel sidecars · openclaw/openclaw@145b57c docs(changelog): omit advisory id from release notes · openclaw/openclaw@d606881 docs(changelog): refresh 2026.5.26 notes · openclaw/openclaw@26c0c19 docs(skills): add release changelog update workflow · openclaw/openclaw@c8d20ae fix(e2e): bound upgrade survivor cli checks · openclaw/openclaw@c965b3a test: speed up doctor config flow tests · openclaw/openclaw@5177180 fix(events): sanitize queued system markers (#87094) · openclaw/openclaw@c1151ea fix(gateway): remove redundant unknown union · openclaw/openclaw@f393ebe test: speed up model fallback tests · openclaw/openclaw@e7f644c fix(imessage): stage remote media before understanding · openclaw/openclaw@ae52be9 fix(gateway): drop stale subagent announce history · openclaw/openclaw@982e888 fix: repair local approval resolution (#86771) · openclaw/openclaw@13cfb77 fix(e2e): bound codex media plugin setup · openclaw/openclaw@f89fcdd fix(gateway): stop chat timeout fallback cascade · openclaw/openclaw@b4f6928 perf(gateway): defer scheduled service imports · openclaw/openclaw@b74cd69 test: speed up capability cli tests · openclaw/openclaw@0126aba perf(gateway): defer startup warning fallback imports · openclaw/openclaw@0ee4ccf fix(gateway): bound watch regression teardown · openclaw/openclaw@7014bd0 fix(codex): keep WebChat delivery hints out of user requests · openclaw/openclaw@0cfccdb
fix(agents): quarantine unsupported tool schemas · openclaw/openclaw@8c644ee
vincentkoc · 2026-05-27 · via Recent Commits to openclaw:main

@@ -194,6 +194,10 @@ import {

194194

} from "../../tool-allowlist-guard.js";

195195

import { UNKNOWN_TOOL_THRESHOLD } from "../../tool-loop-detection.js";

196196

import { normalizeToolName } from "../../tool-policy.js";

197+

import {

198+

filterRuntimeCompatibleTools,

199+

type RuntimeToolSchemaDiagnostic,

200+

} from "../../tool-schema-projection.js";

197201

import {

198202

addClientToolsToToolSearchCatalog,

199203

applyToolSearchCatalog,

@@ -480,6 +484,40 @@ function collectTrustedLocalMediaToolNames(params: {

480484

]);

481485

}

482486487+

function logRuntimeToolSchemaQuarantine(params: {

488+

diagnostics: readonly RuntimeToolSchemaDiagnostic[];

489+

tools: readonly Parameters<typeof getPluginToolMeta>[0][];

490+

runId: string;

491+

sessionKey?: string;

492+

sessionId?: string;

493+

}): void {

494+

if (params.diagnostics.length === 0) {

495+

return;

496+

}

497+

const summary = params.diagnostics

498+

.map((diagnostic) => {

499+

const tool = params.tools[diagnostic.toolIndex];

500+

const pluginId = tool ? getPluginToolMeta(tool)?.pluginId : undefined;

501+

const owner = pluginId ? ` plugin=${pluginId}` : "";

502+

emitTrustedDiagnosticEvent({

503+

type: "tool.execution.blocked",

504+

runId: params.runId,

505+

...(params.sessionKey ? { sessionKey: params.sessionKey } : {}),

506+

...(params.sessionId ? { sessionId: params.sessionId } : {}),

507+

toolName: diagnostic.toolName,

508+

toolSource: pluginId ? "plugin" : "core",

509+

...(pluginId ? { toolOwner: pluginId } : {}),

510+

deniedReason: "unsupported_tool_schema",

511+

reason: diagnostic.violations.join(", "),

512+

});

513+

return `${diagnostic.toolName}${owner}: ${diagnostic.violations.join(", ")}`;

514+

})

515+

.join("; ");

516+

log.warn(

517+

`[tools] quarantined ${params.diagnostics.length} unsupported tool schema${params.diagnostics.length === 1 ? "" : "s"} before model runtime projection: ${summary}. Run openclaw doctor for details.`,

518+

);

519+

}

520+483521

const MAX_BTW_SNAPSHOT_MESSAGES = 100;

484522

const TOOL_SEARCH_CONTROL_ALLOWLIST_NAMES = [

485523

TOOL_SEARCH_CODE_MODE_TOOL_NAME,

@@ -1917,11 +1955,22 @@ export async function runEmbeddedAttempt(

19171955

model: params.model,

19181956

})

19191957

: filteredBundledTools;

1920-

const uncompactedEffectiveTools = filterLocalModelLeanTools({

1958+

const projectedUncompactedEffectiveTools = filterLocalModelLeanTools({

19211959

tools: [...tools, ...normalizedBundledTools],

19221960

config: params.config,

19231961

agentId: sessionAgentId,

19241962

});

1963+

const uncompactedToolSchemaProjection = filterRuntimeCompatibleTools(

1964+

projectedUncompactedEffectiveTools,

1965+

);

1966+

logRuntimeToolSchemaQuarantine({

1967+

diagnostics: uncompactedToolSchemaProjection.diagnostics,

1968+

tools: projectedUncompactedEffectiveTools,

1969+

runId: params.runId,

1970+

sessionKey: params.sessionKey,

1971+

sessionId: params.sessionId,

1972+

});

1973+

const uncompactedEffectiveTools = [...uncompactedToolSchemaProjection.tools];

19251974

let effectiveTools = uncompactedEffectiveTools;

19261975

const catalogToolHookContext = {

19271976

agentId: sessionAgentId,

@@ -1978,11 +2027,20 @@ export async function runEmbeddedAttempt(

19782027

toolHookContext: catalogToolHookContext,

19792028

});

19802029

toolSearchCatalogApplied = true;

1981-

effectiveTools = filterLocalModelLeanTools({

2030+

const projectedToolSearchTools = filterLocalModelLeanTools({

19822031

tools: toolSearch.tools,

19832032

config: params.config,

19842033

agentId: sessionAgentId,

19852034

});

2035+

const toolSearchSchemaProjection = filterRuntimeCompatibleTools(projectedToolSearchTools);

2036+

logRuntimeToolSchemaQuarantine({

2037+

diagnostics: toolSearchSchemaProjection.diagnostics,

2038+

tools: projectedToolSearchTools,

2039+

runId: params.runId,

2040+

sessionKey: params.sessionKey,

2041+

sessionId: params.sessionId,

2042+

});

2043+

effectiveTools = [...toolSearchSchemaProjection.tools];

19862044

if (toolSearch.compacted) {

19872045

prepStages.mark(codeModeControlsEnabledForRun ? "code-mode" : "tool-search");

19882046

log.info(