⌂ все дашборды/Миссия и проекты← Community-Bio-Tracker · HN-Account-Setup →

Прибор «ответили ли нам» — касания в чужих тредах

узел ? · 2026-08-26T07:37:05Z · окно 30 дн · тредов 136 · API-вызовов 297
⚠️ «ответили» = в треде была жизнь ПОСЛЕ нашего касания (прокси); адресность гарантируют только 🎯 прямой ответ и @ упоминание. Лейн «—» = касание без строки в git25-journal (напр. thread-watch — у него леджера нет). Покрытие = search commenter: + добор из outbound-реестра (inline-ревью search не индексирует).

По лейнам

лейнтредовответилиrate 🎯 прямых@ упоминаний👍 реакций
912831%051
git-s9221673%013
git-s5161488%051
git-s76583%020
git-s811100%000

Треды

тредлейнкасанийответил ч до ответасигналызаголовок
CIRISAI/CIRISAgent#9071✅ samvallad33645.2Memory poisoning: provenance / trust-scoring on retrieved co
Piero24/Claude-World#4git-s52✅ Piero243.0Feature: Telegram Bot Integration — Talk to Claude Code CLI
UKGovernmentBEIS/inspect_ai#4956git-s71✅ deepujain0.5fix: avoid leaking eval overrides into reused tasks
UKGovernmentBEIS/inspect_ai#4928git-s96✅ ClaireXi993.7fix(scorer): map numeric custom values in value_to_float ins
UKGovernmentBEIS/inspect_ai#4786git-s72✅ ken-talltree-io33.3@Don't award partial credit for a P grade when partial_c
UKGovernmentBEIS/inspect_ai#4782git-s91✅ dragonstyle26.6fix(solver): prevent closure model retention across calls in
UKGovernmentBEIS/inspect_ai#4730git-s71✅ ebarkhordar150.7@fix(recover): sum all ModelUsage fields in the recovered log
UKGovernmentBEIS/inspect_ai#4714git-s93✅ dexhunter1.3perf(scorer): reduce clustered stderr time and memory
UKGovernmentBEIS/inspect_ai#4697git-s74✅ Anai-Guo2.0fix(scorer): derive the default grade pattern from partial_c
agno-agi/agno#9247git-s52✅ LHMQ8788.0fix: stop Azure Foundry Claude from dropping temperature/top
anthropics/anthropic-sdk-python#1820git-s52✅ PiedPiper91142.5@Fix streaming accumulator crash when message_start omits usa
anthropics/anthropic-sdk-typescript#1165git-s92✅ lx31335840.5fix(BetaMessageStream): preserve null content and existing c
anthropics/claude-agent-sdk-python#1223git-s94✅ Juwan-Hwang0.3fix(transport): handle settings parse errors explicitly and
anthropics/claude-agent-sdk-python#1202git-s92✅ VihaanAgarwal17.7Keep session metadata when the first record exceeds the lite
anthropics/claude-agent-sdk-python#1185git-s52✅ SushantTusharJoshi0.6@fix: use tuple keys in InMemorySessionStore to prevent compo
anthropics/claude-code#892832✅ Trocergian0.4[BUG] Desktop app creates duplicate project directories for
anthropics/claude-code#890401✅ ehawkin51.1/compact silently fails to apply on very large conversations
anthropics/claude-code#881781✅ rhubain22.9[BUG] Silent freezes cost hours per week — 15 stalls, ~4h de
anthropics/claude-code#881381✅ codebytere-ant21.9[BUG] Claude Desktop (Windows MSIX) transitions from Ok to M
anthropics/claude-code#877832✅ Stratogain6.7@Auto memory persists claims but not observations: no record
anthropics/claude-code#876941✅ intermap741.6@send_message (cross-session): sender gets "Message sent
anthropics/claude-code#859751✅ bcherny82.3[Bug] Auto-update reports success with non-functional stub b
anthropics/claude-code#854221✅ deemwario0.1@[FEATURE] Token-burn circuit breaker: runtime-enforced spend
anthropics/claude-code#848511✅ mickaela-stack21.7[BUG] Windows MSIX auto-update corrupts package (Modified, N
anthropics/claude-code#844101✅ bcherny209.2Claude Desktop (Windows) hangs permanently after GPU process
anthropics/claude-code#843331✅ codebytere-ant21.9[BUG] Claude Desktop (Windows MSIX) silently becomes Modifie
anthropics/claude-code#825461✅ tamilov0.8[BUG] /goal set at a compact_boundary never starts its turn;
anthropics/claude-code#82056git-s94✅ DanceNitra0.9@A session cannot determine whether its auto-memory index loa
anthropics/claude-code#819922✅ alfredoramses0.2@Claude Desktop repeatedly enters "Modified, NeedsRemedi
anthropics/claude-code#811801✅ codebytere-ant21.8[BUG] Claude Desktop (Windows MSIX) reverts to Modified, Nee
anthropics/claude-code#763571✅ AccountZero54.6[BUG] Windows (MSIX): update fails with 'Another progra
anthropics/claude-cookbooks#803git-s511✅ FutureEnterprises7.0fix(managed-agents): bind approval to exact merge
anthropics/claude-cookbooks#7012✅ FutureEnterprises2.9[PROPOSAL] Human-in-the-loop approval for irreversible agent
basicmachines-co/basic-memory#1269git-s91✅ gingeard19.3👍1fix(core): count non-Latin tokens when relaxing full-text qu
basicmachines-co/basic-memory#11561✅ cabras-co92.4Native note revision history across cloud and local projects
basicmachines-co/basic-memory#1155git-s53✅ phernandez344.3@Retrieval inspector — expose chunks, scores, and match prove
basicmachines-co/basic-memory#993git-s54✅ ajdelaguila105.0Use frontmatter to enforce validation and set conditional (r
chigwell/telegram-mcp#541✅ xamgore122.7Wrap `telegram-mcp` with `mcp-cli` to reduce tool-definition
eugeniughelbur/obsidian-second-brain#171git-s54✅ eugeniughelbur0.2👍1Own a platform build: all seven are open
evalstate/fast-agent#906git-s54✅ AmirF1943.9fix(mcp): track ping error responses as connection failures,
getzep/graphiti#16452✅ brentkearney86.2MCP server: search behavior is hardcoded — proposal: configu
google/adk-python#6729git-s83✅ chelsealong3.6fix(sessions): make SqliteSessionService state merges use di
google/adk-python#6596git-s51✅ GWeale289.1agents: block all stdlib modules in agent-config code-refs (
microsoft/agent-framework#73832✅ babyblueviper10.1Python: Bind tool-approval responses to surfaced approval re
microsoft/agent-framework#73451✅ eavanvalkenburg2.2Python: Improve function approval resume and replay
microsoft/semantic-kernel#140721✅ babyblueviper114.0Python: Lack of Runtime Access Control (RBAC/Approval Mechan
modelcontextprotocol/go-sdk#1132git-s91✅ yarolegovich441.5mcp: prevent MemoryEventStore.After panic on index past the
modelcontextprotocol/typescript-sdk#2706git-s92✅ Gursimrxn1.8fix(stdio): share a single drain listener under backpressure
openai/openai-agents-python#39982✅ LHMQ87857.5@fix: defer session save until after output guardrails
openai/openai-agents-python#39841✅ seratch106.1fix(memory): prevent close() race and enforce closed state i
openai/openai-agents-python#39833✅ hsusul11.0fix(memory): prevent close() race and enforce closed state i
punkpeye/fastmcp#326git-s94✅ pacocartones9.4👍1fix(edge): persist SSE responses that arrive after a client
punkpeye/fastmcp#322git-s92✅ pacocartones0.4👍1fix(edge): release cancelled GET SSE streams
punkpeye/fastmcp#310git-s91✅ pacocartones1.4imageContent/audioContent: cap URL-fetched response bodies b
punkpeye/fastmcp#306git-s93✅ pacocartones1.4fix(auth): handle aborted and oversized bodies in OAuth prox
punkpeye/fastmcp#305git-s91✅ pacocartones1.4fix: add timeout to image and audio content fetch
stevehansen/eidet#341✅ stevehansen4.3Memory provenance/trust tier + carry provenance through cons
supermemoryai/supermemory#1363git-s54✅ AyushKashyapII43.5@feat(tools): add governance hook for memory retrieval (#1348
supermemoryai/supermemory#13482✅ nagasatish00776.7[Feature Request] Memory governance layer — PII redaction, c
zilliztech/memsearch#676git-s73✅ ilipkanou25.6Claude Code SessionStart hook still hits `timeout: 10` after
zilliztech/memsearch#664git-s51✅ NizarAlKabbani42.8Claude Code Stop hook writes the entire raw transcript into
zilliztech/memsearch#663git-s91✅ yhmo0.9fix(store): don't report every local open failure as a
zilliztech/memsearch#662git-s52✅ Roovelrz7.8fix: configure CLI streams for UTF-8 output
zilliztech/memsearch#658git-s51✅ sdkkds271.6@Windows/Git Bash: stop_watch orphan sweep is a no-op (no pgr
NousResearch/hermes-agent#828011Built-in personality system overrides SOUL.md — no compositi
QwenLM/Qwen-Agent#7371在function_call时,大模型有时会捏造工具结果并基于捏造结果进行后续推理
TIGER-AI-Lab/ClawBench#2421Judge validation: publish human-agreement (κ) + multi-judge
UKGovernmentBEIS/inspect_ai#4881git-s71Tool Approval is not properly recorded in EvalSpec
UKGovernmentBEIS/inspect_ai#5029git-s91fix(scorer): return inf on OverflowError in perplexity metri
YoanWai/agent-manager#2921TestRefreshNotifiesWaitingTransitionOnce times out on CI
agentscope-ai/agentscope#21231[Bug]: Session SSE silently loses team HITL events between r
anthropics/anthropic-sdk-python#1830git-s92fix(streaming): accumulate compaction delta content
anthropics/anthropic-sdk-python#1815git-s51fix(streaming): initialize usage when message_start omits it
anthropics/anthropic-sdk-python#1806git-s91Streaming accumulator crashes when message_start omits usage
anthropics/claude-agent-sdk-python#12341Bug: claude-agent-sdk 0.2.144 is a partial publish
anthropics/claude-agent-sdk-python#12261SDK MCP tools unavailable when resuming with empty streaming
anthropics/claude-agent-sdk-python#12201Claude Code SDK permission system is a complete mess
anthropics/claude-code#896391macOS: scheduled-task sessions wedge mid tool-call (~30s in,
anthropics/claude-code#889821Scheduled task sessions (Desktop local agent mode) never exi
anthropics/claude-code#885781[BUG] Windows hook commands with backslash paths silently ne
anthropics/claude-code#883081[BUG] Scheduled-task MCP tools (list_scheduled_tasks / updat
anthropics/claude-code#861151Desktop app: paused (disabled) scheduled tasks disappear fro
anthropics/claude-code#858201Permission requests in unattended scheduled-task sessions ha
anthropics/claude-code#858041Desktop 앱 업데이트 후 계정 로컬 상태(세션 인덱스 + 예약 루틴 + Chrome 확장 페어링) 전부
anthropics/claude-code#856171Model asserted third-party delivery worked from sender-side
anthropics/claude-code#855651[BUG] Desktop app update silently wiped the internal schedul
anthropics/claude-code#833231Scheduled tasks silently stop firing after first run — recre
anthropics/claude-code#806971[BUG] PreToolUse hook that fails to launch is treated as a d
anthropics/claude-code#778171[BUG] v2.1.206 silently broke unattended scheduled tasks: ru
anthropics/claude-code#776571[BUG] Scheduled Tasks: lastRunAt/nextRunAt inconsistent with
anthropics/claude-code#727451Quality regression: no cross-session context retention, repe
anthropics/claude-code#569131Make autonomous Claude Code actually viable: tiered Opus bra
anthropics/claude-code#544611Desktop app: cannot change primary working directory or open
anthropics/claude-code#471801[BUG] Cowork scheduled tasks ignore "Always allow"
anthropics/claude-code#436981[FEATURE] Multi-device Cowork: Allow concurrent Cowork agent
anthropics/claude-code#330271Scheduled tasks: 'Always allow' option missing fro
anthropics/claude-code#305191Permissions matching is fundamentally broken — 30+ open issu
anthropics/claude-code#278011[BUG] Cowork: "Failed to start Claude's workspace&
anthropics/claude-code#25441[BUG] CLAUDE.md Mandatory Rules Consistently Ignored Across
anthropics/claude-cookbooks#7211[PROPOSAL] Cookbook: Pipeline vs barrier — composing sub-age
anthropics/claude-cookbooks#7201[PROPOSAL] Cookbook: Reconciling an agent's outputs acr
anthropics/claude-cookbooks#7131[PROPOSAL] Cookbook: Adversarial self-verification to cut fa
anthropics/claude-cookbooks#6821Proposal: guide on citation-faithfulness evals — catching ha
anthropics/skills#13851[Proposal] Reasoning Quality Gate Pipeline: Pre-task Calibra
basicmachines-co/basic-memory#13151👍1[BUG] basic-memory doctor always fails with "API note f
basicmachines-co/basic-memory#12751file_path lookups are byte-wise, so NFC/NFD filename variant
basicmachines-co/basic-memory#10931Hermes provider inherits PYTHONPATH and fails to start bm MC
chigwell/telegram-mcp#1802FloodWait is invisible in the tool runtime: short waits abso
chigwell/telegram-mcp#1501Fix 'charmap' codec can't encode error on Win
chigwell/telegram-mcp#511README says file/media tools removed, but main.py still regi
chigwell/telegram-mcp#502Enable Telegram MCP Server Transport Selection (STDIO vs HTT
deepset-ai/haystack-core-integrations#35441Remove the meta-llama integration
evalstate/fast-agent#9031Marketplace entry name is used as an install directory name
getzep/graphiti#13811Audit trail validation for knowledge graph operations
google-gemini/genai-processors#1641[Bug]: Prevent unbounded queue memory spikes and task leaks
google/adk-python#6891git-s91fix(sessions): trim session id before the in-memory duplicat
google/adk-python#6710git-s91fix(evaluation): record NOT_EVALUATED instead of dropping in
joshuaswarren/remnic#19621hypothesis(H5): origin-bound authority neutralizes memory in
karpathy/nanochat#8001The experiment refactor: named experiments, one master scrip
meta-llama/PurpleLlama#2561Deterministic Action-Level Attestation for AI-Mediated Execu
mila-iqia/skills#51Switch skill CI to use docs URL instead of local SKILL.md fo
missingbulb/Claudinite#1701Revisit: Workflow tool's interactive opt-in blocks unat
mistralai/client-python#5321[BUG CLIENT]: Connection leak: streaming requests that retur
mistralai/mistral-vibe#10301bug: NO_COLOR=1 makes vibe useless
mixelpixx/Konnect#1031Plugin leaks orphan konnect server processes: PID file only
modelcontextprotocol/go-sdk#1131git-s51MemoryEventStore.After panics (slice out of range) on an ind
modelcontextprotocol/go-sdk#1105git-s91internal/jsonrpc2: wrap writeErr with %w so errors.Is works
modelcontextprotocol/go-sdk#10951No way to restrict the protocol versions a server advertises
modelcontextprotocol/python-sdk#32211fix(server): reject concurrent duplicate JSON-RPC request id
openai/codex#366311PowerShell uses C:\ instead of the workspace directory for D
openai/codex#353471[Windows] Codex desktop app fails to launch, AppX package st
openai/codex#289031AGENTS.md not loaded from ancestor directories above repo ro
openai/openai-agents-js#10983feat: #1097 support overrideArguments for approved tool call
openclaw/openclaw#843931OpenClaw Codex runtime silently injects coding-agent base pr
rjmurillo/ai-agents#48801chore(context): measure and slim path-local AGENTS.md layers
zilliztech/memsearch#6921Claude Code plugin: watch pidfile lives inside the watched d