Skip to content
Guidance/Daily Briefing
digesteveryonedigestevidenceradarmodelagent7 min read · Aug 5, 2026

Radar brief — 2026-08-05

Security patches and session control dominate: Goose ships a critical CLI fix, Claude Code hardens sandbox credentials, and Codex/Cline push deeper multi-thread and checkpoint workflows.

What the evidence shows

Versioned GitHub releases lead this window (Claude Code, Codex, Goose, Cline, OpenCode). Enterprise auth, model routing, and agent session UX are the recurring moves; two OpenCode “PR evidence” tags are noise.

Claude Codev2.1.221
Key Signal & Impact
VS Code Focus view hides tool noise; Linux/WSL sandbox mode: "mask" for credential files (macOS falls back to deny) — tighter secrets handling in agent runs
OpenAI Codex CLIrust-v0.146.0
Key Signal & Impact
Named/pinned sessions, thread forks, Agent Plugins + Bedrock/Claude marketplaces, remote Code Mode over WebSocket — multi-conversation agent ops
OpenAI Codex CLIrust-v0.145.0
Key Signal & Impact
Paginated thread history, /import from Cursor & Claude Code (MCP/plugins/sessions), experimental Bedrock + GPT-5.6 Sol default
OpenAI Codex CLIrust-v0.144.6
Key Signal & Impact
Bundled GPT-5.6 Sol/Terra/Luna prompts; context windows corrected to 272k
Google Antigravity1.1.10
Key Signal & Impact
Business sign-in for Gemini Enterprise (Cloud project, license seats, regional inference, admin controls)
Goosev1.44.0
Key Signal & Impact
Security: arbitrary command execution via goose review / git core.fsmonitor (GHSA-r5pp-p5r8-466r); plus latest Gemini models, Stop hook working_dir, Sakana Fugu provider
Clinedesktop-v0.0.8
Key Signal & Impact
Edit any earlier message → fork + atomic workspace rewind; reliable checkpoints after restart/compaction
Clinecli-v3.0.49
Key Signal & Impact
/undo fixed after tool use; checkpoints recorded again; full workspace rewind (respects .gitignore)
Clinesdk/sdk/v0.0.67
Key Signal & Impact
Reasoning effort/budget from models.dev catalog; Anthropic thinking modes; OpenRouter default → anthropic/claude-sonnet-5
OpenCodev1.18.5
Key Signal & Impact
Claude adaptive thinking, Mistral reasoning/cache stability, MiniMax thinking variant, server terminal transport fixes
Gemini CLIv0.54.0-preview.0 / v0.55.0-nightly.*
Key Signal & Impact
Prerelease: session ID rotate on model fallback, HTTPS enforcement for Google creds, capacity exhaustion as terminal, Antigravity agent runner work
xAI Grok CLI (Grok Build)ed6d543
Key Signal & Impact
ACP session resume/close, cap live subagents at 16, model switch during plan approval, MCP image extract before truncation, sandbox deny-glob fix
Windsurfv3.6.27
Key Signal & Impact
Windows cert store for Devin Desktop (corp TLS proxies); local edit tools refuse writes through symlinks
Zed AIv1.13.1
Key Signal & Impact
Branch picker filtering/grouping; runnable gutter run-status; image-viewer zoom — IDE polish more than agent core

For developers

  • Upgrade Goose to v1.44.0 now if you use goose review or shared git config — the GHSA is arbitrary command execution via core.fsmonitor.
  • Claude Code v2.1.221: try Focus view (Ctrl+Alt+F) on long tool-heavy turns; on Linux/WSL, prefer sandbox credential mode: "mask" over broad secret exposure.
  • Codex CLI: if you juggle side quests, use rust-v0.146.0 session pin/fork and /new; migrating from Cursor/Claude — rust-v0.145.0 /import pulls MCP, plugins, and memories.
  • Cline users: desktop-v0.0.8 + cli-v3.0.49 restore real checkpoints and /undo after tool use; treat edit-earlier-message as a full workspace rewind, not chat-only.
  • OpenCode v1.18.5: worth it if you hit Claude thinking shape bugs, Mistral cache drift, or broken server terminal transport.
  • Windsurf v3.6.27: required behind corporate TLS inspection on Windows; symlink write refusal is a real foot-gun fix for local agents.
  • Grok Build (ed6d543): note the 16 live subagent cap and model-switch-during-plan-approval when scripting multi-agent flows.
  • Skip Gemini CLI nightlies/previews for daily drivers unless you are chasing specific core fixes (HTTPS creds, capacity hang).

For teams

  • Security patch lane: Goose GHSA + Windsurf symlink/cert fixes + Claude sandbox masking — schedule these before feature rollouts.
  • Enterprise Google path: Antigravity 1.1.10 Business sign-in ties Gemini Enterprise seats, region, and admin controls — align IdP/Cloud project policy before wide deploy.
  • Codex as migration hub: rust-v0.145–0.146 supports Bedrock, plugin marketplaces (including Claude Code), and Cursor/Claude import — useful for standardizing MCP/skills across a mixed fleet.
  • Cline checkpoint model: transactional workspace rewind changes incident recovery and “try again from step N” workflows; document that restores touch the filesystem, not only the transcript.
  • Model defaults shifting: Cline SDK OpenRouter → Claude Sonnet 5; Codex Bedrock default GPT-5.6 Sol and 272k context metadata — update cost/latency budgets and allowlists.
  • Open proposals (P5): team-facing improve/review items exist for Codex 0.145/0.144.6, Cline, Goose, OpenCode, Grok, and Zed pre — triage if those assistants are in your standard image.

What to ignore

  • OpenCode pr-38252-videos and pr-37967-screenshots-final — erification artifact tags, not product releases.
  • Gemini CLI stacked nightlies/previews — useful signal on direction (Antigravity runner, caretaker, stream errors), not stable version pins.
  • Zed AI v1.13.1 agent tags are light; the release notes are mostly IDE UX (branch picker, gutters), not a model/agent step-change.
  • Raw “+N more changes” counts without the linked security or session semantics above.

Watch next

  • Goose post-GHSA follow-ups and whether goose review / git integration hardens further.
  • Codex plugin marketplace + remote Code Mode adoption; stability of Bedrock/GPT-5.6 Sol defaults.
  • Cline checkpoint/fork behavior under long multi-agent runs (desktop + CLI + SDK alignment).
  • Claude Code sandbox mask coverage beyond Linux/WSL and Focus view as default team UX.
  • Gemini CLI promotion of preview/nightly fixes (HTTPS creds, capacity terminal, session rotate) into a stable tag.
  • Windsurf/Devin Local symlink and cert patterns as a template for other desktop agents on corp networks.
  • Open proposals converting from P5 backlog into accepted upgrade paths for Codex, Goose, Cline, OpenCode.

How to use DevAgentRadar

  • Read What the evidence shows first — every row is a citable version, not a rumor.
  • Developers: apply the concrete upgrade and flag notes; prefer GitHub release tags over changelog marketing pages.
  • Teams: use the security and defaults section to sequence patches, then migrations (import/plugins/enterprise auth).
  • Ignore prerelease artifact tags and low-signal IDE polish unless they match your stack.
  • Recheck Watch next and open proposals when you refresh standards docs or golden images.
  • This brief is judgment from the supplied evidence window only — no invented releases or scores.

Evidence appendix

  • Evidence window: top curated signals (versioned releases preferred)
  • Assistants tracked: 22
  • Signals in this brief: 18
  • Trusted source mix (GitHub/changelog vs HTML): 55%
  • Open improvement proposals: 8
  • Sources reporting errors: 0

High-signal releases used

  • Claude Code v2.1.221: [VSCode] · Added Focus view: a chat-menu toggle that hides tool activity behind an expandable per-turn summary with a live running-tool indicator, toggled with — source

  • OpenAI Codex CLI rust-v0.146.0: Name new sessions with /new or /clear, pin important threads, and switch between side conversations without closing them. (#34605, #34840, #35011) · Support — source

  • Google Antigravity 1.1.10: Added Business sign-in for Gemini Enterprise accounts, so you can authenticate with a Google Cloud project under Google Cloud terms, use a license seat allocate — source

  • OpenAI Codex CLI rust-v0.145.0: Added experimental paginated thread history with efficient resume, search, persisted names, sub-agent support, and memories. (#33364, #33907, #34085, #34229, #3 — source

  • Goose v1.44.0: Arbitrary command execution in goose CLI via goose review (via git core.fsmonitor config) - [GHSA-r5pp-p5r8-466r](https://github.com/aaif-goose/goose/securitysource

  • Zed AI v1.13.1: This week's release includes improved branch picker filtering and grouping, run-status indicators for runnable gutter controls, and editable image-viewer zoom l — source

  • Cline desktop-v0.0.8: Edit any earlier message in a conversation — the app forks the session at that point, rewinds the workspace to that run's checkpoint, and re-runs from your edit — source

  • Cline sdk/sdk/v0.0.67 — SDK v0.0.67: Reasoning controls (effort, budget, on/off) are now driven by the models.dev catalog and normalized once before provider encoding, so requests match what each p — source

  • Cline cli-v3.0.49: /undo works again once the agent has used tools — the checkpoint picker counted tool results as user turns, so restore aborted with "Could not find user messa — source

  • OpenCode v1.18.5: Improve Claude adaptive thinking handling across more response shapes. · Avoid OpenAI Responses phase handling that could break some conversations. · Preserve g — source

  • Gemini CLI v0.54.0-preview.0: Changelog for v0.53.0-preview.0 · Changelog for v0.52.0 · chore(release): bump version to 0.54.0-nightly.20260722.gf743ab579 · fix(caretaker): sanitize and wrap — source

  • Gemini CLI v0.55.0-nightly.20260729.g3499c84f7: chore/release: bump version to 0.54.0-nightly.20260728.gbef611950 · feat(pr-generator-db): implement Firestore concurrency dual-locking and test ingestion utili — source

  • Gemini CLI v0.55.0-nightly.20260801.gf47d6c6f7: fix(core): classify capacity exhaustion as terminal to prevent retry hangs · fix(core,cli): propagate InvalidStreamError details to UI for specific empty respon — source

  • OpenAI Codex CLI rust-v0.144.6: Refreshed bundled instructions for GPT-5.6 Sol, Terra, and Luna, and corrected their context windows to 272,000 tokens. (#33972, #34009) · #33972 Backport refre — source

Open proposals

Was this guidance useful?

Anonymous · one vote per visitor · re-click to clear. Helps improve digests—not a product ranking.