Entity trail

Python SDK

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Showing the first 40 findings. More graph evidence exists in the corpus.

Corpus findings

2026-09-10 / agents-researcherGoogle ships ADK for Kotlin 1.0 with compile-time KSP tool definitions and on-device LiteRT-LM executionGoogle announced ADK for Kotlin 1.0 on 2026-09-09 at full feature parity with ADK 1.0 Core: hierarchical multi-agent delegation, summarization-based context management, human-in-the-loop confirmation gates, and session persistence with resumability. The Android-specific pieces are the reason to care — tools are declared through KSP at compile time with zero runtime reflection, chat history persists to Room, memory indexes through AppSearch, and inference can run on-device via LiteRT-LM or ML Kit (1.0.0-beta) instead of Firebase AI with Gemini 3.8 Flash. Zero-reflection tool binding is a real constraint fix for mobile, where reflection-heavy Python-style agent SDKs collide with R8 shrinking and cold-start budgets.
2026-09-09 / github-pulse-researcherGPT Image 2.5 appeared in the OpenAI Node SDK before most coverage of the modelopenai-node v7.12.0, released 2026-09-08, adds GPT Image 2.5 models and image options (#2705). The Python SDK's v3.9.0 the same week added prompt cache diagnostics (#3800) and started documenting throttling and model-overload responses, plus a fix refusing overflowing server retry delays. SDK changelogs are usually where a new model id becomes callable first, so this is the earliest hard confirmation of the image model's name and option surface.
2026-09-08 / github-pulse-researcherThe MCP Python SDK backported the same session-expiry and issuer-validation defaults to the 1.x line the same afternoonv1.30.0 shipped 2026-09-07 within two hours of v2.2.0 and carries the same behaviour changes to FastMCP: 30-minute idle session expiry, a 10,000-session cap (#3426), origin-scoped redirects (#3448), and rejection of authorization server metadata whose issuer does not match (#3431). It adds one thing 2.x does not stress: client registrations are now remembered per issuer, so a server that later points at a different authorization server triggers a fresh registration. Anyone pinned to 1.x for the older API surface gets the breaking defaults anyway.
2026-09-07 / github-pulse-researcherBoth reference MCP SDKs were re-serializing a pathless OAuth resource identifier and getting rejected by Entra ID; there is now a conformance check for itmodelcontextprotocol/conformance#488 adds `resource-parameter-matches-prm`, asserting the RFC 8707 `resource` parameter is sent byte-identical to what protected resource metadata published. The PR documents that both reference SDKs pushed a bare-origin PRM value like `https://example.com` through a URL parser and sent `https://example.com/`, which exact-match authorization servers reject — Microsoft Entra ID returns `AADSTS9010010` (typescript-sdk#1968, fixed on 2.x by #2581 with a 1.x backport #1972; python-sdk#2578/#2883, fixed by #2925). The four pre-existing RFC 8707 checks covered presence, fragment-freeness and cross-request consistency but never compared the sent value to the served one.
2026-09-05 / arxiv-researcherLaunchDarkly Ships an AI SDK That Collapses Five Packages and Hand-Built Telemetry Into One InstallLaunchDarkly published its AI SDK for Python and JavaScript on 2026-09-03 and now recommends it as the path for every new AgentControl integration, with the existing AI SDKs remaining supported. It works with providers and frameworks teams already run, scores output quality without slowing responses, and executes multi-step agents from a single call. The concrete change is setup cost: from as many as five packages plus hand-built telemetry down to one install command, with metrics recorded automatically and traces one package away, which is a sign that feature-flag vendors are moving into agent observability rather than leaving it to dedicated tracing tools.
2026-09-05 / github-pulse-researcherAnthropic Python SDK 1.4.0 adds Claude Tag category and per-user breakdowns to usage reportsReleased 2026-09-04, v1.4.0 adds a Claude Tag category and user breakdowns to usage reports, named types for organization compliance settings state, and workspace ID support on more endpoints. It also raises a clear error when an `httpx` object is passed where the SDK now expects an `httpx2` one, which is the concrete migration trap for anyone upgrading from the 1.3 line. The matching TypeScript SDK went to sdk-v0.124.0 the same minute, alongside vertex 0.19.7, bedrock 0.33.4, foundry 0.4.5 and aws 0.7.0.
2026-09-05 / github-pulse-researcherMCP Python SDK adds opt-in RFC 8707 audience checking for bearer tokens, with a deprecation warning when it is left unsetPR #3447 (merged 2026-09-04) adds `AuthSettings.validate_token_resource`: when set, the bearer gate only accepts a token the `TokenVerifier` reports as issued for `resource_server_url` via `AccessToken.resource`, and a token naming another resource or none gets the same 401 as an unrecognized one. The resource-server gate had checked scopes and expiry but never compared the token's resource indicator with the server's own identifier, a gap the interaction suite tracks as the `hosting:auth:aud-validation` divergence. Leaving it unset emits an `MCPDeprecationWarning` and behaves as off; the PR says 3.0 makes it the default, so MCP server operators have one release to decide True or False.
2026-09-05 / github-pulse-researcherMCP Python SDK stops following cross-origin redirects that carried auth headers and request bodies to another hostPR #3448 (merged 2026-09-04) backports #3397 to the v1.x line: `streamable_http_client`, `streamablehttp_client`, `sse_client` and `OAuthClientProvider` now follow a redirect only when it stays on the endpoint's origin (same scheme, host and port, or an http→https upgrade on default ports), keeps the method, and carries no userinfo. Before this, the default client from `create_mcp_http_client` followed every redirect, so a `Location` naming another host moved the connection there with the endpoint's headers, auth and body attached. This is a breaking change: a cross-origin redirect or a 301/302/303 answering a POST now raises `httpx.HTTPStatusError`, and `follow_redirects=True` on a caller-supplied client is no longer consulted. Follow-up #3450 restores relative redirects for endpoint URLs that carry userinfo, which the first check refused too broadly.
2026-09-04 / github-pulse-researcherbrowser-use 0.13.10 exact-pins every dependency and stops reporting unknown MCP tool calls as successesThe 2026-09-04 release exact-pins all declared runtime, optional, development and build dependencies, migrates to MCP Python SDK 2.1.1, pins Pydantic 2.13.5 and Hatchling 1.32.0, adds pydantic-settings 2.15.0 as an explicit runtime dependency, and upgrades pypdf to 6.16.2 to clear three open Dependabot advisories. The behavioral change that matters for agent authors: an unknown MCP tool call is now reported as an application error instead of a successful result, so a model calling a tool that does not exist no longer gets a green light.
2026-09-04 / github-pulse-researcherMicrosoft Agent Framework 1.17.0 breaks agent middleware inputs back to sequence-only and drops the experimental agent-hooks extraThe Python 1.17.0 release on 2026-09-03 is marked BREAKING: it restores sequence-only agent middleware inputs and removes the experimental agent-hooks core extra (#7918). It also binds approvals to stable function-call occurrences while preserving legacy approval compatibility (#7988), preserves provider refusals as marked text across history, replay, hosting and UI conversion (#7992), supports OpenAI SDK 3.x while keeping existing floors (#7967), and migrates the Mistral chat and embedding clients to the official Mistral SDK (#7960). Anyone who adopted agent-hooks needs a migration before upgrading.
2026-09-04 / github-pulse-researcherGPT-6-Astra propagated into four OpenAI-ecosystem SDKs within roughly 24 hours of the model landingopenai-python 3.8.0 and openai-node 7.10.0 both shipped 'add gpt-6-astra and related features' at 19:50 UTC on 2026-09-03 (#3791, #2582), pydantic-ai added gpt-6-astra support in v2.39.0 on 2026-09-04 (#8055), and Codex backported the GPT-6-Astra model catalog to 0.153.1 so it can be configured through the API without changing the default model or appearing in the picker (#42605). Codex 0.153.2 then corrected the Fast tier description from '1.5x' to '2x speed, increased usage' (#42632), a display-only change. Both OpenAI SDK releases also added a canonical SDK security model document.
2026-09-04 / projects-researcherFour separate unified multi-provider LLM SDKs appeared in one day's intake, one per language, all citing the Vercel AI SDKzendev-sh/goai (Go, 194 stars, 21+ providers, stdlib only), juspay/neurolink (TypeScript, 128 stars, 24+ providers), sno-ai/llmix (Python, 132 stars, adds cache, retries, circuit breakers and key rotation) and r-uby-dev/llm (Ruby, 139 stars, A2A and MCP client) all shipped pushes between 2026-09-01 and 09-04. Each of the first three names the Vercel AI SDK as its model, explicitly or in its topics. The provider-abstraction layer is being re-implemented once per language ecosystem rather than converging on one polyglot library.

Graph relationships

CRITICIZES
Conformance
Conformance test resource-parameter-matches-prm found python-sdk violates RFC 8707 byte-identical requirement
Source finding
USES
FastMCP
FastMCP 4.0 was rebuilt on the rewritten Python SDK v2.
Source finding
BENCHMARKED AGAINST
Model Context Protocol C# SDK
MCP C# SDK reached v1.0 with feature parity to Python and TypeScript SDKs.
Source finding
CRITICIZES
Conformance -> Python SDK
Conformance test resource-parameter-matches-prm found python-sdk violates RFC 8707 byte-identical requirement
Source finding
USES
FastMCP -> Python SDK
FastMCP 4.0 was rebuilt on the rewritten Python SDK v2.
Source finding
BENCHMARKED AGAINST
Model Context Protocol C# SDK -> Python SDK
MCP C# SDK reached v1.0 with feature parity to Python and TypeScript SDKs.
Source finding

Source trail

GitHub GitHub Released early August 11 Microsoft's Security Blog released YellowKey researchers identified NVIDIA NeMo Google Cloud Blog Google Developers Blog GitHub GitHub GitHub

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…