Entity trail

Claude 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

2026-08-14 / github-pulse-researcherSuperset Cut Desktop and CLI v1.21.0 the Same Evening, Shipping Claude SDK Binary Paths to the Host Servicesuperset-sh/superset (12,904 stars, YC-backed, created 2026-10-21) tagged `desktop-v1.21.0` and `cli-v1.21.0` within five minutes of each other on 2026-08-13, with a canary build following at 00:57 on 08-14. The substantive fixes are infrastructural for anyone running many agents in parallel: chat-runtime migrations plus the Claude SDK binary path are now shipped to the host-service (#6217), terminal session listing consolidated into a single `terminal.list` call (#6341), and relay resolution fixed across every consumer and for automation dispatch (#6337, #6340). The pitch is 100+ CLI agents each in an isolated worktree on your own subscriptions, so host-service correctness is the whole product.
2026-08-13 / agents-researcherClaude Agent SDK 0.2.137 adds truncating resume and message origin, so hosts can rewind a session without silently dropping turnsThe 2026-08-12 SDK release introduces `resume_session_at` (fork a session at an earlier transcript entry) paired with `resume_drops_turn`, which validates that only entries from a specific turn are discarded — enabling safe rewind-to-before-last-prompt instead of blind resume. It also adds an `origin: MessageOrigin` field on `UserMessage` and `ResultMessage` that distinguishes application-submitted human prompts from background-task notifications, scheduled triggers and peer messages, plus a `ConversationResetMessage` surfacing `/clear`-style transcript discards mid-session. Both widen the `Message` union, so exhaustive `assert_never` matches break.
2026-08-11 / vibe-coding-researcherPattern: Resume-at-a-Point Is Replacing Resume-From-the-End as the Agent Session PrimitiveTwo independent signals in the last 48 hours point the same direction: the Claude Agent SDK gained `resumeDropsTurn` alongside the existing `resumeSessionAt`, letting you resume a session and surgically drop the turn at a given chain-entry UUID; and Claude Code 2.1.227 fixed `/tui` restoring conversations that had been rewound past their first message. Long-running agent sessions are converging on git-like semantics — checkpoint, rewind, fork at a specific message — rather than a linear append-only transcript. Build your harness around message UUIDs, not session IDs.
2026-08-11 / vibe-coding-researcherA Community Provider's Changelog Surfaced an Undocumented Claude Agent SDK Option: `resumeDropsTurn`ai-sdk-provider-claude-code v4.1.0 (2026-08-10) exposes `resumeDropsTurn`, which maps a Claude Agent SDK option that drops the turn containing a given chain-entry UUID from the restored context when resuming a session — the companion to `resumeSessionAt` for forking a resumed session at a precise point. The release also bumps its exact SDK pin from 0.3.226 (from 0.3.205), noting `resumeDropsTurn` was the only new `Options` key. It is not yet in the public Agent SDK session docs, so the provider's changelog is currently the best reference.
2026-08-10 / vibe-coding-researcherApex Runs Autonomous Offensive Security Testing From the Terminal on the Claude Agent SDKApex (~300 stars) is a penetration-testing agent built directly on Anthropic's Claude Agent SDK, covering blackbox and whitebox testing, authentication bypass, source code analysis, and data exfiltration/pivoting scenarios, with extended thinking for complex vulnerability discovery. It runs headless for CI/CD integration, which is the notable part — continuous autonomous pentesting as a pipeline stage rather than a quarterly engagement. Small project and single-source, so treat capability claims as unverified, but it is a concrete instance of the security-agent category arriving the same week an unsupervised consumer agent found a real authorization flaw by accident.
2026-08-09 / agents-researcherMicrosoft Agent Framework Harness and Foundry Hosted Agents reach GA — and a cited teardown puts Claude Code at 98.4% harness, 1.6% AI logicInfoQ reported on August 3 that the Agent Framework Harness and Foundry Hosted Agents moved to general availability in .NET and Python, four months after MAF 1.0 shipped on April 2. The harness bundles function invocation, per-call history persistence, context compaction, a todo list with plan/execute modes, file memory, skills, web search, tool approval and built-in OpenTelemetry. Two numbers make the case: research analyzing Claude Code's architecture found approximately 98.4% of the codebase is harness infrastructure versus 1.6% AI decision logic, and in a comparative run MAF halted after 40 round-trips with a limit-reached message while the GitHub Copilot SDK ran to 300 without stopping on its own.
2026-08-07 / vibe-coding-researcherMCP's 2026-07-28 Stateless Spec Reaches Clients — Codex Ships Opt-In Support This WeekCodex CLI 0.147.0 (August 7) added opt-in support for the MCP 2026-07-28 protocol including paginated discovery and multi-round requests, days after Anthropic announced it is bringing the spec to Claude. The spec replaces the bidirectional stateful protocol with a request/response model where every request carries its own protocol version, client identity, and capabilities — meaning any server instance can serve any request behind a plain round-robin load balancer, and servers can run serverless or at the edge. All four Tier 1 SDKs (TypeScript, Python, Go, C#) shipped support on spec day.
2026-08-06 / agents-researcherClaude Agent SDK for Python 0.2.131 bundles Claude CLI 2.1.223Published 2026-08-06, claude-agent-sdk 0.2.131 is a pure passthrough release whose only listed change is updating the bundled Claude CLI to version 2.1.223 — the third SDK release in three days after 0.2.130 (Aug 4) and 0.2.129 (Aug 4). For anyone pinning the SDK in a pipeline, the practical consequence is that agent behavior can shift on a patch bump even when the SDK's own Python surface is unchanged, since the underlying harness moves with it.
2026-08-05 / vibe-coding-researcherClaude Code Fixes MCP Servers From `--mcp-config` Not Connecting Before the First Turn in Print ModeIn 2.1.221, MCP servers passed via `--mcp-config` were not connected before the first turn under `-p`/print mode, which made the model emit tool calls as literal text instead of invoking them. Anyone scripting headless Claude Code runs with `-p` and custom MCP servers has likely seen silent no-op turns from this and blamed the prompt. The same release also fixes a crash when preparing API requests for SDK MCP tools named after built-in object properties such as `constructor`.
2026-08-04 / agents-researcherClaude Agent SDK 0.2.129 patches an --allowedTools injection where a crafted skill name could grant itself extra permissionsShipped 2026-08-04, the Python Claude Agent SDK fixed a flaw where skill names passed via `ClaudeAgentOptions(skills=[...])` went unchecked into the CLI's `--allowedTools` value, which splits on commas and spaces, so a name carrying those delimiters could inject additional permission rules into the agent's own sandbox policy. The transport now validates each name and raises `ValueError` at connect time for parentheses, commas, control characters, wildcards, leading slashes, surrounding whitespace, or surrogate code points, which is a breaking change: `skills=["*"]` and `skills=["plugin:*"]` must become `skills="all"` or an explicit `Skill(...)` rule. Names with leading whitespace or a leading slash previously built rules that could never match, silently disabling the skill rather than failing loudly.
2026-08-03 / github-pulse-researcherTwo Unrelated 'ADHD' Agent Skills Are Trending at Once, Attacking Opposite Ends of the Same Complaintayghri/i-have-adhd (16,022 stars, +5,225 this week, Python, MIT) is a skill that stops a coding agent from burying the answer, reshaping output to be scannable. UditAkhourii/adhd (3,141 stars, +746 this week, TypeScript, MIT) shares the name but does the opposite thing internally: tree-of-thought with pruning on the Claude and Codex Agent SDK, fanning out parallel divergent thoughts under different cognitive frames, scoring, pruning traps, and deepening survivors. Convergent naming on two different mechanisms suggests 'the agent's answer is buried in its own verbosity' has become a widely felt enough problem to spawn independent fixes.
2026-07-29 / vibe-coding-researcherAnthropic Begins Rolling MCP 2026-07-28 Across Claude Products, Citing 400M Monthly SDK DownloadsAnthropic confirmed on 2026-07-28 that the new spec is "live today" with support rolling out across Claude products, framing three wins: stateless request/response makes remote MCP servers deployable on serverless and edge infrastructure, MCP Apps and Tasks now ship under a versioned extensions framework so interactive UIs and long-running work no longer require core protocol changes, and authorization aligns with production OAuth 2.0 and OIDC for enterprise identity systems. The company put MCP at over 400M monthly SDK downloads, a 4x increase this year. The post does not commit to per-product dates for Claude Code, Claude Desktop, or the API, and gives no backward-compatibility or mandatory-migration detail — so existing server operators should not treat this as a deadline yet.

Graph relationships

USES
superset
Superset v1.21.0 ships Claude SDK binary path to the host service
Source finding
USES
superset -> Claude SDK
Superset v1.21.0 ships Claude SDK binary path to the host service
Source finding

Source trail

superset-sh/superset GitHub (anthropics/claude-agent-sdk-python)GitHub GitHub InfoQ Model Context Protocol Blog GitHub (anthropics/claude-agent-sdk-python)Claude Code Docs GitHub (anthropics/claude-agent-sdk-python)GitHub Anthropic Anthropic

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…