Who contributed to Anthropic on May 14, 2026?

4 developers shipped this update, including rlancemartin, iainmcgin, yordis, and tejas-dharani.

What were the notable Anthropic updates?

feat(managed_agents): Linear stateless webhook bridge template, feat(managed_agents): CMA Sessions API as an MCP server (stdio + HTTP), and fix(buffa): closed-enum JSON helpers decode via Enumeration, not Deserialize.

@anthropics

Claude SDKs and developer tooling

github ↗

Pick a date

Topics: AI / ML Full archive →

The Wire · Showcase

MANAGED AGENTS BREAK INTO LINEAR AND CLAUDE DESKTOP — BUFFA CLOSES JSON SERIALIZATION GAPS

By RepoJournal · Filed 06:02 UTC on May 14, 2026 · About Anthropic

4 people shipped this

rlancemartin @rlancemartin 2 cited

iainmcgin @iainmcgin 2 cited

yordis @yordis 1 cited

tejas-dharani @tejas-dharani 1 cited

Two production-ready bridges just shipped: Linear workflows now trigger Claude Managed Agents via stateless webhooks, and Claude Desktop users can chat with hosted CMA agents directly through an MCP server.

The Linear integration [1] is the story here — a minimal TypeScript/Bun template that connects Linear's Agent Platform to Claude Managed Agents using CMA outbound webhooks with zero session state management. When someone @mentions the agent in Linear, it spins up a CMA session, runs inference to idle on Anthropic's servers, and posts the result back to Linear. No held SSE streams, no database mapping — just metadata in the session object. The companion piece [2] makes this bidirectional: MCP server wrapping the Sessions API means Claude Desktop and claude.ai can both invoke hosted CMA agents via tool calls and wait for results, solving the "how do I publish my agent" problem until that becomes first-class. Over on the Rust side, buffa closed three critical gaps in protobuf JSON handling. The closed-enum deserializers [3] were incorrectly bounded on `DeserializeOwned`, forcing enum types to implement Serialize when codegen only emits Enumeration — now they decode directly via the Enumeration trait. The extern-path routing [4] fixes descriptor.proto imports by routing `google.protobuf` prefixes to the right types, and ancillary file emission [5] now respects the documented contract: empty companion files are dead code and no longer clutter the tree. View serialization [6] rounds it out — `FooView<'_>` types can now JSON-serialize without allocating a full owned message, which defeats the entire purpose of views. The Claude Code and Agent SDK ecosystem bumped to 2.1.141 [7] [8] with corresponding action syncs across the CI/CD surface [9].

Action items

→ Deploy Linear webhook template if you're running Linear + Managed Agents — this is production-ready anthropics/claude-cookbooks [plan]
→ If you built custom MCP servers for CMA, compare against the new stdlib wrapper — consolidate if it covers your use case anthropics/claude-cookbooks [monitor]
→ Rebuild buffa-generated code against the latest codegen — closed-enum and view serialization fixes are live anthropics/buffa [plan]

References

[1] feat(managed_agents): Linear stateless webhook bridge template ↗ anthropics/claude-cookbooks
[2] feat(managed_agents): CMA Sessions API as an MCP server (stdio + HTTP) ↗ anthropics/claude-cookbooks
[3] fix(buffa): closed-enum JSON helpers decode via Enumeration, not Deserialize ↗ anthropics/buffa
[4] feat(buffa-codegen): file-level extern path routing for descriptor.proto ↗ anthropics/buffa
[5] fix(buffa-codegen): omit empty ancillary content files ↗ anthropics/buffa
[6] view: impl Serialize for generated view types (#83) ↗ anthropics/buffa
[7] chore: bump Claude Code to 2.1.141 and Agent SDK to 0.2.141 anthropics/claude-code-action
[8] chore: bump bundled CLI version to 2.1.141 anthropics/claude-agent-sdk-python
[9] Sync from claude-code-action base-action@86eb26b anthropics/claude-code-base-action

Quick answers

What shipped in Anthropic on May 14, 2026?: Two production-ready bridges just shipped: Linear workflows now trigger Claude Managed Agents via stateless webhooks, and Claude Desktop users can chat with hosted CMA agents directly through an MCP server. In total, 12 commits, 7 pull requests, and 1 releases landed.
Who contributed to Anthropic on May 14, 2026?: 4 developers shipped this update, including rlancemartin, iainmcgin, yordis, and tejas-dharani.
What were the notable Anthropic updates?: feat(managed_agents): Linear stateless webhook bridge template, feat(managed_agents): CMA Sessions API as an MCP server (stdio + HTTP), and fix(buffa): closed-enum JSON helpers decode via Enumeration, not Deserialize.

CLAUDE AGENT SDK FIXES STDIN CLOSURE BUG IN BACKGROUND TASKS

The Python SDK's query method was prematurely closing stdin when background subagents were still in flight, breaking tool calls and silently bypassing hooks - now fixed in v0.2.127.

ai 16 shipped 1-min read

@anthropics 1 day ago

BUFFA v0.9.1 PATCHES REGRESSION THAT BROKE ORDINARY SCHEMAS

Buffa shipped an emergency patch overnight after v0.9.0 broke codegen on schemas with a few hundred proto files, small enough to hit in production.

ai 113 shipped 1-min read

@anthropics 2 days ago

CLAUDE FOUNDATION MODELS ADDS APP ATTEST, PLUGIN ECOSYSTEM STABILIZES

iOS developers can now ship Claude calls without API keys in the binary, backed by Apple's Secure Enclave.

ai 111 shipped 1-min read

@anthropics 3 days ago

BUFFA FIXES V0.9.0 REGRESSION, CLAUDE PLUGINS ECOSYSTEM GAINS 9 NEW INTEGRATIONS

Buffa's codegen was silently rejecting moderately large schemas due to a memory bound meant only for untrusted input, but a fix shipped overnight restores throughput and clears the path for googleapis-scale proto sets.

ai 156 shipped 2-min read

Elsewhere on the wire

AI Agents about 9 hours ago

CLAUDE OPUS 5 LANDS ACROSS THE STACK

The newest Anthropic model is now live in langchain, Cline, and llama-index, with native support for extended reasoning and 1M context windows.

ai-agents 28 shipped 1-min read

Local LLMs about 9 hours ago

OLLAMA LANDS LAGUNA SUPPORT AND CRUSHES MEMORY LEAKS WHILE SGLANG HITS V0.5.16 WITH CONFIDENCE-DRIVEN SPECULATIVE DECODING

Ollama shipped three critical performance and reliability fixes for Metal residency and concurrent access patterns, while SGL-Lang released 0.5.16 with a new speculative algorithm hitting 383.7 tok/s on DeepSeek-V4.