Skip to content

Release v3.3.5

Choose a tag to compare

@github-actions github-actions released this 02 Sep 20:44
· 91 commits to main since this release

What's Changed

  • fix(v1): resolve the installation reference before connecting a Checks repo by @chelojimenez in #4600
  • fix(github-checks): re-list repositories when the bindings change by @chelojimenez in #4601
  • WebMCP viewport: sharp at rest, honest geometry, and a stream that answers a slow link by @chelojimenez in #4592
  • feat(models): offer the GPT-5.6 family to BYOK OpenAI keys by @nachocossio in #4454
  • Add GPT-5.6 BYOK models to changelog by @mintlify[bot] in #4611
  • Codex over app-server: a transport that can actually pause by @chelojimenez in #4595
  • Fix/bb 63 swarm server readiness by @weimar-torres-herrera in #4503
  • Document every harness id the builtin-tools route accepts by @chelojimenez in #4603
  • docs: Claude Code approval works with MCP servers; fix the Codex refusal reason by @chelojimenez in #4613
  • Cursor harness: accept a brokered credential, unblocking hosted evals and swarms by @chelojimenez in #4605
  • Playground: a reopened conversation stops describing itself with your current settings by @chelojimenez in #4606
  • v1 chat: host targeting, so an agent can drive a harness — or be told it can't by @chelojimenez in #4608
  • Playground: record the environment a conversation actually ran in by @chelojimenez in #4609
  • v1 errors, composed credential grants, and a skew probe by @chelojimenez in #4610
  • cursor/auto is a sentinel, not a provider — and it is what a Cursor session records by @chelojimenez in #4607
  • Docs: Cursor CLI host model display and misconfigured host error by @mintlify[bot] in #4622
  • Local Claude Code: run the real agent natively on the user's machine by @chelojimenez in #4604
  • github checks: remediation labels, the feedback-comment toggle, and docs by @chelojimenez in #4621
  • sdk: register eval_iteration as a permalink resource type by @chelojimenez in #4625
  • Lead the Evaluate run page with what broke and what to do about it by @chelojimenez in #4627
  • models: brand logos for 11 catalog providers, and wire spacexai to xAI by @ZeHuari in #4439
  • swarms: keep the thread of a running swarm (BB-74) by @ZeHuari in #4394
  • fix(swarms): stop the live strip contradicting its own session count by @nachocossio in #4392
  • Swarm: complete UI revamp -> Findings tab, Create flow, Insights, Persona revamp by @prathmeshpatel in #4372
  • chore: add release changeset for inspector, CLI, and SDK by @ignaciojimenezr in #4630
  • fix(playground): connect a server without leaving the Playground (BB-134) by @nachocossio in #4501
  • Show a swarm's own name as its Overview title by @nachocossio in #4460
  • Say what the Evaluate run launched against, and show the grading pair by @chelojimenez in #4632
  • ci(release): skip the local-harness pack until its digests are committed by @chelojimenez in #4633
  • Gate member-only surfaces on the Convex actor, not the WorkOS user object by @olartgabo in #4505
  • chore: add release changeset for inspector, CLI, and SDK by @ignaciojimenezr in #4636

Full Changelog: v3.3.4...v3.3.5