| Репо | Issue | Заголовок | Наш актив (PR-порт) | Матч | Мерж (liveness) | Возраст | Флаги |
|---|---|---|---|---|---|---|---|
| microsoft/agent-framework | #8698 | Python: [Feature]: Roadmap for AnyIO-backed asyncio and Trio support | Pipeline-vs-barrier concurrency (замер 1.93×) ↗ Verified-ops SDK agent (verify before act) ↗ | L | HIGH | 1д | запрос контентаавтор=мейнт |
| openai/openai-agents-python | #5085 | Agent.as_tool() nested subagent runs never hit prompt cache, even with byte-identical instructions/tools/promp | Pipeline-vs-barrier concurrency (замер 1.93×) ↗ Verified-ops SDK agent (verify before act) ↗ | L | HIGH | 6д | запрос контента |
| microsoft/agent-framework | #8503 | .NET: [Bug]: Rejected tool approvals are re-requested unless CreateResponse(false, reason) provides a reason ( | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ | L | HIGH | 6д | запрос контента |
| google/adk-python-community | #171 | feat(plugins): add AuthorityRoutingPlugin — pre-execution authority posture (ADVISE/EXECUTE/DEFER/STOP) | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ | L | HIGH | 63д | запрос контентачужой PR |
| google/adk-python-community | #169 | feat(memory): add MongoDBMemoryService with Atlas Vector Search support | Persistent memory / second-brain (context resets) ↗ | L | HIGH | 66д | запрос контента |
| google/adk-python-community | #113 | Add SidClaw Governance Service integration | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ MCP tool trust / verification before execution ↗ | L | HIGH | 176д | запрос контентачужой PR |
| google/adk-python-community | #99 | feat(memory): add DatabaseMemoryService with SQL backend and agent scratchpad | Persistent memory / second-brain (context resets) ↗ | L | HIGH | 196д | запрос контента |
| google/adk-python-community | #69 | Feature Request: Add Milvus-backed Memory Service | Persistent memory / second-brain (context resets) ↗ | L | HIGH | 226д | запрос контента |
| google-gemini/cookbook | #1389 | Support `excludeDomains` for `google_search` grounding in the Gemini Developer API (currently Vertex/Enterpris | Citation-faithfulness eval (hallucinated/unsupported citations) ↗ | L | MED-HIGH | 1д | запрос контента |
| google-gemini/cookbook | #1353 | Feature Request/Cookbook Proposal: Safety Guardrails and Real-Time Fact-Verification Workflow with Gemini API | Citation-faithfulness eval (hallucinated/unsupported citations) ↗ Adversarial self-verification / burden-of-proof ↗ Multi-agent consensus / voting ↗ Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ MCP tool trust / verification before execution ↗ Verified-ops SDK agent (verify before act) ↗ | L | MED-HIGH | 25д | запрос контентачужой PR |
| microsoft/semantic-kernel | #14412 | Python: CVE-2026-26030 / GHSA-xjw9-4gw8-4rqx: question on the PR:L component of the CVSS vector | Pipeline-vs-barrier concurrency (замер 1.93×) ↗ Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ Persistent memory / second-brain (context resets) ↗ Verified-ops SDK agent (verify before act) ↗ | L | MED | 13д | запрос контентачужой PR |
| openai/openai-cookbook | #3062 | [PROBLEM] False-positive "distillation" ban on Pro subscriber - automated appeal rejected by bot, no human rev | Adversarial self-verification / burden-of-proof ↗ Multi-agent consensus / voting ↗ Persistent memory / second-brain (context resets) ↗ Pipeline-vs-barrier concurrency (замер 1.93×) ↗ Verified-ops SDK agent (verify before act) ↗ | L | MED | 19д | запрос контента |
| microsoft/semantic-kernel | #14302 | Feature Request: Native Token Metering & Bounded Overshoot (x402 Protocol) | Verified-ops SDK agent (verify before act) ↗ Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ | L | MED | 37д | запрос контента |
| microsoft/semantic-kernel | #14062 | Python: [Feature] Memory validation layer to protect against memory poisoning attacks (OWASP AMG integration) | Persistent memory / second-brain (context resets) ↗ Adversarial self-verification / burden-of-proof ↗ MCP tool trust / verification before execution ↗ | L | MED | 107д | запрос контента |
| GoogleCloudPlatform/generative-ai | #2904 | [Bug]: Computer use example has been updated to gemini-3.5-flash but the computer use tool does not support th | Computer-use / GUI agent ↗ Pipeline-vs-barrier concurrency (замер 1.93×) ↗ | L | MED | 115д | запрос контента |
| microsoft/semantic-kernel | #14047 | Python: [Feature Request] Memory Poisoning Protection for Semantic Kernel via OWASP Agent Memory Guard | Persistent memory / second-brain (context resets) ↗ Adversarial self-verification / burden-of-proof ↗ MCP tool trust / verification before execution ↗ | L | MED | 115д | запрос контента |
| microsoft/semantic-kernel | #14032 | Python: Feature: Add MCP server trust verification for agent tool safety | MCP tool trust / verification before execution ↗ | L | MED | 123д | запрос контента |
| microsoft/semantic-kernel | #14025 | .Net: C#: Feature request: Memory poisoning defense (OWASP ASI06) for Semantic Kernel memory stores | Persistent memory / second-brain (context resets) ↗ | L | MED | 127д | запрос контента |
| meta-llama/PurpleLlama | #230 | Add AgentThreatBench to CyberSecEval: OWASP Agentic Top 10 benchmark (UK AISI) | Adversarial self-verification / burden-of-proof ↗ Persistent memory / second-brain (context resets) ↗ MCP tool trust / verification before execution ↗ | L | MED | 128д | запрос контентачужой PR |
| meta-llama/PurpleLlama | #231 | AgentThreatBench: OWASP Agentic Top 10 benchmark — complement to CyberSecEval for agentic threat scenarios | Persistent memory / second-brain (context resets) ↗ Adversarial self-verification / burden-of-proof ↗ MCP tool trust / verification before execution ↗ | L | MED | 128д | запрос контента |
| microsoft/semantic-kernel | #14011 | Python: Feature request: Memory poisoning defense (OWASP ASI06) integration | Persistent memory / second-brain (context resets) ↗ Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ MCP tool trust / verification before execution ↗ | L | MED | 133д | запрос контента |
| meta-llama/PurpleLlama | #147 | How to combine PromptGuard-2 and AlignmentCheck? | Adversarial self-verification / burden-of-proof ↗ MCP tool trust / verification before execution ↗ | L | MED | 298д | запрос контента |
| GoogleCloudPlatform/generative-ai | #2272 | [Bug]: MemoryBank with Crewai example not working | Persistent memory / second-brain (context resets) ↗ | L | MED | 400д | запрос контента |
| GoogleCloudPlatform/generative-ai | #2046 | [Bug]: discoveryengine.SearchRequest can't find results | Citation-faithfulness eval (hallucinated/unsupported citations) ↗ | L | MED | 501д | запрос контента |
| GoogleCloudPlatform/generative-ai | #1204 | [Bug]: evaluate_gemini_with_autosxs: my computer SvcAcct doesnt have access to my bucket. | Verified-ops SDK agent (verify before act) ↗ Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ | L | MED | 722д | запрос контента |
| microsoft/agent-framework | #8678 | Python: [Bug]: BedrockChatClient silently drops guardrailConfig and other Bedrock options, so GPT-6 Sol/Luna/A | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ | M | HIGH | 1д | запрос контентачужой PR |
| microsoft/agent-framework | #8707 | .NET: Foundry hosting rejects Harness local chat history marker when AllowStoredOutputEnabled is false | Persistent memory / second-brain (context resets) ↗ | M | HIGH | 1д | запрос контентачужой PR |
| microsoft/agent-framework | #8703 | Python: [Feature]: Make public `run` and `get_response` methods async in vNext | Pipeline-vs-barrier concurrency (замер 1.93×) ↗ | M | HIGH | 1д | запрос контентаавтор=мейнт |
| microsoft/agent-framework | #8700 | Python: [Bug]: Mixed approval and Host results give a tool result the user role | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ Verified-ops SDK agent (verify before act) ↗ | M | HIGH | 1д | запрос контента |
| microsoft/agent-framework | #8674 | .NET: [Bug]: AIAgent.RunStreamingAsync does not surface the final image from HostedImageGenerationTool | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ | M | HIGH | 1д | запрос контента |
| openai/openai-agents-python | #5157 | `ShellTool` and `ApplyPatchTool` default `needs_approval=False` for host-side execution | | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ | M | HIGH | 2д | запрос контента |
| microsoft/agent-framework | #8632 | Python: raw-data content mappings lose annotations and attachment metadata | Citation-faithfulness eval (hallucinated/unsupported citations) ↗ | M | HIGH | 2д | чужой PR |
| microsoft/agent-framework | #8663 | Python: [Bug]: LocalShellTool never executes on a self-hosted Foundry agent since openai 1.14.4 - "No tool out | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ | M | HIGH | 2д | запрос контентачужой PR |
| microsoft/agent-framework | #8638 | .NET: Proposal: agent-framework-spider, a web crawl/scrape/search tool package | Persistent memory / second-brain (context resets) ↗ MCP tool trust / verification before execution ↗ | M | HIGH | 2д | запрос контента |
| microsoft/agent-framework | #8633 | Python: [Bug]: Oversized MessagePack history records make sessions unreadable | Verified-ops SDK agent (verify before act) ↗ | M | HIGH | 2д | запрос контентачужой PR |
| microsoft/agent-framework | #8627 | .NET: [Bug]: StateManager is not synchronized, but InProcessRunner runs a superstep's executors concurrently — | Pipeline-vs-barrier concurrency (замер 1.93×) ↗ | M | HIGH | 2д | чужой PR |
| microsoft/agent-framework | #8574 | Python: Design safe provider-outbox recovery for completed mixed pauses | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ | M | HIGH | 3д | запрос контентаавтор=мейнт |
| microsoft/agent-framework | #8617 | Python: One client per vendor, or one that needs no credential? | Pipeline-vs-barrier concurrency (замер 1.93×) ↗ | M | HIGH | 3д | запрос контента |
| microsoft/agent-framework | #8577 | .NET: Handoff workflow raises two approval requests for one tool call, and resuming reports an error after the | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ | M | HIGH | 3д | запрос контента |
| microsoft/agent-framework | #8559 | .NET: Python: Add OpenTelemetry spans for agent, chat, and function middleware | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ | M | HIGH | 4д | запрос контента |
| openai/openai-agents-python | #5088 | Modal sandbox: evaluate native filesystem APIs and clarify overwrite compatibility | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ | M | HIGH | 6д | запрос контента |
| microsoft/agent-framework | #8441 | .NET: [Bug]: Opening user message is dropped from InMemoryChatHistoryProvider when compaction is enabled and t | Persistent memory / second-brain (context resets) ↗ | M | HIGH | 7д | запрос контентачужой PR |
| openai/openai-agents-python | #5054 | Cancelling a MCP server work task does not stop the worker | MCP tool trust / verification before execution ↗ | M | HIGH | 8д | запрос контентачужой PR |
| openai/openai-agents-js | #1934 | getAllMcpTools only rejects duplicate tool names across MCP servers, so two tools on the same server that norm | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ MCP tool trust / verification before execution ↗ | M | HIGH | 9д | запрос контентачужой PR |
| microsoft/agent-framework | #8407 | .NET: [Bug]: Hosted-agent runtime never activates OTLP export to a third-party destination, despite docs descr | MCP tool trust / verification before execution ↗ | M | HIGH | 9д | запрос контента |
| openai/openai-agents-python | #5005 | EncryptedSession.pop_item drains recoverable ciphertext when configured with the wrong key | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ Persistent memory / second-brain (context resets) ↗ Pipeline-vs-barrier concurrency (замер 1.93×) ↗ | M | HIGH | 10д | запрос контентачужой PR |
| microsoft/agent-framework | #8292 | Python: [Bug]: Functional workflow checkpoint replay can swap cached results between concurrent calls to the s | Pipeline-vs-barrier concurrency (замер 1.93×) ↗ Persistent memory / second-brain (context resets) ↗ | M | HIGH | 13д | запрос контента |
| google/adk-python-community | #183 | feat(plugins): optional SeenRelay ambient measurement for repeated read-only tool calls | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ Persistent memory / second-brain (context resets) ↗ | M | HIGH | 15д | запрос контента |
| openai/openai-agents-python | #4827 | Deferred interrupted-turn session items are never persisted when an approval resume goes through next_step_run | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ Persistent memory / second-brain (context resets) ↗ Verified-ops SDK agent (verify before act) ↗ | M | HIGH | 22д | запрос контентачужой PR |
| openai/openai-agents-js | #1775 | code-change-verification cannot run on Windows under Node 20.12.2 or newer | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ Pipeline-vs-barrier concurrency (замер 1.93×) ↗ | M | HIGH | 26д | запрос контентачужой PR |
| openai/openai-agents-python | #4740 | SQLAlchemySession(create_tables=True) cannot create tables on MySQL due to unbounded String columns | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ Persistent memory / second-brain (context resets) ↗ | M | HIGH | 27д | запрос контентачужой PR |
| microsoft/agent-framework | #7831 | Python: [Bug]: get_latest returns a stale checkpoint when the newest one cannot be decoded | Pipeline-vs-barrier concurrency (замер 1.93×) ↗ Verified-ops SDK agent (verify before act) ↗ | M | HIGH | 32д | запрос контента |
| google/adk-python-community | #177 | Proposal: fail-closed AgentFuse gate in `before_tool_callback` | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ Persistent memory / second-brain (context resets) ↗ | M | HIGH | 37д | запрос контента |
| google/adk-python-community | #155 | feat: Add ValkeyMemoryService with vector similarity search (Valkey Search module) | Adversarial self-verification / burden-of-proof ↗ Persistent memory / second-brain (context resets) ↗ | M | HIGH | 112д | запрос контентачужой PR |
| openai/openai-agents-js | #1353 | Firefox WebRTC Realtime sessions disconnect mid-turn; likely ICE consent / USE-CANDIDATE interop | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ | M | HIGH | 126д | запрос контента |
| google/adk-python-community | #142 | feat: post-execution audit trail pattern to complement AgentGovernancePlugin | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ | M | HIGH | 127д | запрос контента |
| google/adk-python-community | #129 | [Community Project] create-google-adk-agent — one-command scaffold for ADK projects | Pipeline-vs-barrier concurrency (замер 1.93×) ↗ Multi-agent consensus / voting ↗ Persistent memory / second-brain (context resets) ↗ MCP tool trust / verification before execution ↗ | M | HIGH | 140д | |
| microsoft/agent-framework | #4842 | .NET: Discussion: Agent Identity and Trust for Multi-Agent Workflows | Multi-agent consensus / voting ↗ Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ Pipeline-vs-barrier concurrency (замер 1.93×) ↗ Verified-ops SDK agent (verify before act) ↗ | M | HIGH | 186д | запрос контента |
| google/adk-python-community | #103 | feat: add FirestoreSessionService for serverless session persistence | Persistent memory / second-brain (context resets) ↗ | M | HIGH | 187д | запрос контента |
| google/adk-python-community | #97 | Integration Proposal: SDC Agents — Deterministic Semantic Data Artifacts from Enterprise Datastores | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ | M | HIGH | 197д | запрос контента |
| google/adk-python-community | #91 | feat: Add LangExtract tool for structured information extraction | Citation-faithfulness eval (hallucinated/unsupported citations) ↗ | M | HIGH | 217д | запрос контентачужой PR |
| google/adk-python-community | #88 | feat(plugins): add fallback plugin to automatically fallback between Gemini models | Persistent memory / second-brain (context resets) ↗ | M | HIGH | 218д | запрос контента |
| google/adk-python-community | #46 | Add self-hosted option to manage working memory (sessions) longterm memory management with ADK | Persistent memory / second-brain (context resets) ↗ | M | HIGH | 240д | чужой PR |
| google/adk-python-community | #7 | feat: add Typesense-based session service | Persistent memory / second-brain (context resets) ↗ | M | HIGH | 339д | запрос контента |
| openai/openai-agents-js | #585 | Realtime Agent Using `function_call` Instead of `mcp_call` for Hosted MCP Tools | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ MCP tool trust / verification before execution ↗ | M | HIGH | 343д | запрос контента |
| google/adk-python-community | #21 | First-class Temporal integration for Google ADK | Multi-agent consensus / voting ↗ Pipeline-vs-barrier concurrency (замер 1.93×) ↗ | M | HIGH | 384д | запрос контента |
| huggingface/cookbook | #325 | Invalid Notebooks | Citation-faithfulness eval (hallucinated/unsupported citations) ↗ | M | HIGH | 385д | запрос контента |
| huggingface/cookbook | #316 | Execution Error in Agents recipes | Multi-agent consensus / voting ↗ | M | HIGH | 431д | запрос контента |
| huggingface/cookbook | #303 | Contribution to Hugging Face 🤗 cookbook: Add a search agent cookbook | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ | M | HIGH | 481д | запрос контентачужой PR |
| huggingface/cookbook | #90 | ValidationError: 1 validation error for agenerate | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ Pipeline-vs-barrier concurrency (замер 1.93×) ↗ | M | HIGH | 870д | запрос контента |
| openai/openai-agents-python | #5162 | OpenAIResponsesCompactionSession: compacted items fail replay with "Unknown parameter: 'input[N].created_by'" | Persistent memory / second-brain (context resets) ↗ Pipeline-vs-barrier concurrency (замер 1.93×) ↗ Verified-ops SDK agent (verify before act) ↗ | M | HIGH | 0д | запрос контента |
| microsoft/agent-framework | #8711 | Python: Support workflow entry-state updates and per-AgentExecutor run options | Authority routing ADVISE / EXECUTE / DEFER / STOP ↗ | M | HIGH | 0д | запрос контентаавтор=мейнт |
| microsoft/agent-framework | #8724 | Python: [Feature]: No sample or documentation for grounding an agent on MongoDB | Citation-faithfulness eval (hallucinated/unsupported citations) ↗ Persistent memory / second-brain (context resets) ↗ Verified-ops SDK agent (verify before act) ↗ | M | HIGH | 0д | запрос контентачужой PR |
| google-gemini/genai-processors | #164 | [Bug]: Prevent unbounded queue memory spikes and task leaks in `streams.concat` | Pipeline-vs-barrier concurrency (замер 1.93×) ↗ Persistent memory / second-brain (context resets) ↗ | M | MED-HIGH | 78д | чужой PR |
| google-gemini/cookbook | #1274 | google_search grounding silently disabled when structured/JSON output is requested (gemini-3.5-flash) | Citation-faithfulness eval (hallucinated/unsupported citations) ↗ Persistent memory / second-brain (context resets) ↗ | M | MED-HIGH | 94д | запрос контента |
| Issue | Заголовок | Наш матч | Возраст | Обновл. | 💬 | Флаги |
|---|---|---|---|---|---|---|
| #325 | Invalid Notebooks | citation-faithfulness | 385д | 385д | 0 | |
| #316 | Execution Error in Agents recipes | consensus | 431д | 431д | 0 | |
| #303 | Contribution to Hugging Face 🤗 cookbook: Add a search agent cookbook | authority-routing | 481д | 62д | 4 | чужой PR |
| #90 | ValidationError: 1 validation error for agenerate | authority-routing, pipeline-barrier | 870д | 870д | 0 | |
| #370 | Recipe idea: catching LLM-as-judge bias in agent evals with open models | 8д | 8д | 0 | ||
| #363 | "Post training a VLM for reasoning with GRPO using TRL" Not Work | 74д | 74д | 0 | ||
| #339 | Add Quantum Circuit Optimization with Qiskit and VQE Notebook | 285д | 285д | 0 | ||
| #337 | 404 on https://api-inference.huggingface.co/models/bigcode/starcoder | 325д | 325д | 0 | ||
| #328 | An Error In "Fine-Tuning a Vision Language Model (Qwen2-VL-7B) with the Hugging Face Ecosystem (TRL)" Notebook | 376д | 376д | 0 | ||
| #326 | Improve the Knowledge Graphs cookbook | 380д | 380д | 1 | ||
| #322 | "Building RAG with Custom Unstructured Data" - SimpleLocalConfig | 417д | 417д | 0 | ||
| #313 | 🧑🍳 New Cookbook Recipe: Add a recipe showcasing the usage of Pruna with Diffusers | 436д | 435д | 4 | ||
| #310 | Invalid Notebooks | 444д | 444д | 0 | чужой PR | |
| #301 | Bring Arabic to the Hugging Face Cookbook! | 482д | 482д | 1 | ||
| #292 | Non generic example causes. Cannot copy out of meta tensor; no data! Please use torch.nn.Module.to_empty() ins | 550д | 550д | 0 | ||
| #291 | mongodb_smolagents_multi_micro_agents.ipynb | 561д | 561д | 1 | ||
| #268 | Reusing index with from_index() caused "No passages provided" | 626д | 626д | 0 | ||
| #216 | Turkish translations | 703д | 80д | 14 | мейнт | |
| #185 | Weird redirection in URL in `Advanced RAG on Hugging Face documentation using LangChain` cookbook | 755д | 755д | 0 | мейнт | |
| #183 | "RAG with unstructured data", uses `documents` instead of `docs` / unused `docs` variable? | 760д | 760д | 0 | ||
| #181 | Incorrect links to the source files | 763д | 763д | 0 | ||
| #92 | Translate to Russian (RU) | 867д | 866д | 0 | ||
| #83 | Quantization Stable Diffusion | 882д | 882д | 0 | чужой PR | |
| #82 | Call for Contributions | 883д | 772д | 15 | мейнт | |
| #72 | Structured Text-to-SQL generation using Hugging Face and Outlines | 903д | 154д | 2 | ||
| #65 | Train | 915д | 913д | 2 | ||
| #57 | RAG: Question Answering using Gemma, Elasticsearch & langchain | 933д | 913д | 1 | ||
| #48 | Some outputs of code cells don't show up on hf.co/learn | 940д | 451д | 1 | ||
| #43 | Comprehensive Cookbook for Fine-tuning Gemma Model on Mental Health Assistant Dataset | 945д | 940д | 4 | ||
| #39 | Contribution to Hugging Face 🤗 cookbook: Add a LangChain agent that can interact with a graph database | 948д | 947д | 1 | ||
| #34 | Translate to Simplified Chinese (zh-CN) | 949д | 660д | 4 | ||
| #25 | Idea for contribution : Recipe on Training FineTuning Embeddings / Re-Rankers on Custom Domain Specific Datase | 951д | 951д | 2 |