Skip to content
Guidance/Daily Briefing
digesteveryonedigestevidenceradaragentmodel7 min read · Aug 7, 2026

Radar brief — 2026-08-07

Security patches and agent-control surfaces dominate: Goose ships a critical CLI advisory, Codex tightens cyber-model auto-review, Cline hard-blocks plan-mode edits, and Claude Code plus Codex push self-hosted runners and portable plugins.

What the evidence shows

Versioned GitHub releases carry most of the weight this window (16 signals across 22 tracked assistants; ~55% GitHub/changelog mix). Strongest clusters: agent plugin/runtime control, plan/approval security, and enterprise auth.

Claude Codev2.1.224
Key Signal & Impact
Self-hosted runner for web/mobile/desktop sessions (Team/Enterprise); archive plugin install over HTTPS with optional SHA-256 pin; Bedrock region-prefix env; sandbox/permissions/compaction surfaces touched
OpenAI Codex CLIrust-v0.147.0
Key Signal & Impact
Portable Agent Plugins + multi-catalog search; --approve-for-me auto-reviewed approvals; Cursor skill import; conversation sections; sandbox/--full-auto tagged
OpenAI Codex CLIrust-v0.146.0
Key Signal & Impact
Plugin manifests, workspace publishing, Bedrock/Claude Code marketplaces; named sessions, thread fork (incl. temporary); remote Code Mode over WebSocket
OpenAI Codex CLIrust-v0.146.1
Key Signal & Impact
Backport: safer automatic-review defaults for cyber-capable models; clearer permission-change UX
Clinev4.1.4
Key Signal & Impact
Plan mode no longer model-switched; hard-block of file-editing shell (redirects, mutating git, package installs); skills shown with workflows in slash menu
Clinedesktop-v0.0.8
Key Signal & Impact
Edit-any-earlier-message forks session, rewinds workspace checkpoint transactionally; reliable checkpoints after restart/compaction/resume
Clinesdk/v0.0.67
Key Signal & Impact
Reasoning effort/budget normalized from models.dev catalog; Anthropic thinking modes handled explicitly; OpenRouter default → anthropic/claude-sonnet-5
Goosev1.44.0
Key Signal & Impact
Security: GHSA-r5pp-p5r8-466r — arbitrary command execution via goose review / git core.fsmonitor; Gemini model support; Stop hook gets working_dir
Google Antigravity1.1.10
Key Signal & Impact
Business sign-in for Gemini Enterprise (Cloud project, license seats, regional inference, admin controls)
Gemini CLIv0.54.0
Key Signal & Impact
Enforce HTTPS on GoogleCredentialsAuthProvider (cleartext leak fix); rotate session ID on model fallback; caretaker sanitization
OpenCodev1.18.5
Key Signal & Impact
Claude adaptive-thinking handling; Mistral reasoning history + prompt-cache stability; correct cache keys per SDK
Zed AIv1.15.0-pre
Key Signal & Impact
git.diff_base (HEAD vs merge-base); Project Panel drag-out; JSX/TSX linked editing + Emmet in arrow/return bodies (prerelease)
Windsurfv3.6.27
Key Signal & Impact
Devin Local tools refuse writes through symlinks; Windows cert-store load fixed for TLS-inspecting proxies (HTML changelog)
xAI Grok CLIa5589e9
Key Signal & Impact
Sandbox plan/metadata/repos manifest types; /resume search fix; permission-body collapse; 5xx retry — commit sync, not a semver tag

Low-confidence / noisy: Gemini v0.54.0-preview.0 and v0.55.0-nightly.* overlap changelog noise with v0.54.0; Grok is a monorepo sync hash; Windsurf is a marketing changelog page, not a GitHub release tag.

For developers

  • Patch Goose now if you run CLI ≤ prior to v1.44.0 — treat GHSA-r5pp-p5r8-466r as urgent; avoid goose review until upgraded.
  • Codex: try Agent Plugins and --approve-for-me on rust-v0.147.0; if you stay on 0.146.x, at least take 0.146.1 for cyber-model auto-review defaults.
  • Cline plan mode: v4.1.4 changes the contract — you drive plan→act; don’t expect the model to flip modes, and don’t rely on shell file mutations while planning.
  • Cline Desktop: use message-edit + checkpoint rewind (desktop-v0.0.8) for safe prompt experiments; restores are workspace-atomic.
  • Claude Code: pin plugins via archive + SHA-256 when you can’t use git/npm; set ANTHROPIC_BEDROCK_REGION_PREFIX if cross-region Bedrock routing matters.
  • Gemini CLI: prefer v0.54.0 over nightlies for the HTTPS credential fix and session-id rotation on fallback.
  • OpenCode v1.18.5: worth grabbing if you hit Claude thinking-shape breaks or Mistral cache/history drift.
  • Windsurf/Devin Local: symlink write refusal is a real foot-gun fix — retest any flows that edited through links.

For teams

  • Enterprise auth path: Antigravity 1.1.10 Business sign-in maps Gemini Enterprise seats, regions, and admin controls — pilot under a dedicated Cloud project before org-wide rollout.
  • Claude Code self-hosted runner (Team/Enterprise): treat machines/containers as an execution plane — define image hardening, network egress, and who can attach web/mobile/desktop sessions.
  • Codex plugin marketplaces (0.146–0.147): decide allowed catalogs (local/personal/workspace/remote) and publishing rules before developers wire Bedrock/Claude Code plugin sources.
  • Approval policy: standardize on Codex --approve-for-me vs manual review; 0.146.1’s safer cyber-model defaults should be the floor for higher-risk models.
  • Cline: roll v4.1.4+ so plan mode is policy-enforced (hard blocks), not prompt-enforced; pair with Desktop checkpoints for auditable rewinds.
  • Goose: mandate v1.44.0+ org-wide; add advisory GHSA-r5pp-p5r8-466r to your vuln tracker and verify no shared git core.fsmonitor abuse path remains.
  • Proxy/corp CA: Windsurf v3.6.27 restores Windows cert-store trust — needed behind TLS inspection.

What to ignore

  • Gemini nightly/preview tags that mostly restate the same caretaker/release chore noise as v0.54.0.
  • Zed v1.15.0-pre for production standards — useful IDE signal, still prerelease.
  • Raw Grok monorepo sync commits until they land a clear version tag.
  • Inflated “+N more changes” tails without the named security/agent control items above.
  • Open proposals still at P5 backlog (older Codex 0.144/0.145, Cline 4.0.7, etc.) — not evidence of new releases.

Watch next

  • Follow-up fixes or CVSS detail after Goose GHSA-r5pp-p5r8-466r.
  • Codex plugin ecosystem adoption post-rust-v0.147.0 (catalog trust + --approve-for-me defaults in CI).
  • Claude Code self-hosted-runner operational patterns on Team/Enterprise.
  • Whether Cline’s plan-mode hard blocks and Desktop atomic rewind spread to SDK/CLI consistently past sdk/v0.0.67 and v4.1.4.
  • Zed v1.15.0 stable notes for any MCP/agent breaks hinted by prerelease tags.
  • Open proposals: Goose v1.44.0, OpenCode v1.18.5, Codex 0.145/0.144.6 improvements.

How to use DevAgentRadar

  • Read What the evidence shows first — versioned tables only, no vendor spin.
  • Developers: apply concrete CLI/IDE actions in your own toolchain today.
  • Teams: turn the same signals into policy (approvals, plugin sources, self-hosted runners, advisory SLAs).
  • Ignore prerelease noise and unversioned marketing; Watch next for the next evidence gate.
  • Improve the radar via open proposals when a release deserves deeper team or developer follow-up.

Evidence appendix

  • Evidence window: top curated signals (versioned releases preferred)
  • Assistants tracked: 22
  • Signals in this brief: 16
  • Trusted source mix (GitHub/changelog vs HTML): 55%
  • Open improvement proposals: 8
  • Sources reporting errors: 0

High-signal releases used

  • Claude Code v2.1.224: Added self-hosted environments: claude self-hosted-runner turns your own machines or containers into a place Claude Code web, mobile, and desktop sessions can — source

  • OpenAI Codex CLI rust-v0.147.0: Install portable Agent Plugins and search across local, personal, workspace, and remote plugin catalogs. (#36544, #36409, #36919, #36796) · Organize conversatio — 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

  • Zed AI v1.15.0-pre: This week's release includes a new git.diff_base setting for choosing whether editor gutters, file-status colors, and git: diff show uncommitted changes aga — 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

  • 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

  • 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/security — source

  • Cline v4.1.4: Recognize Chutes as a provider. · Show skills alongside workflows in the slash command menu, and disambiguate commands that share a name instead of letting one — source

  • Cline sdk/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

  • OpenAI Codex CLI rust-v0.146.1: Apply safer automatic-review defaults for cyber-capable models and explain permission changes in the terminal interface. (#37057) · #37057 [0.146] Backport safe — source

  • Gemini CLI v0.54.0: 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

  • 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

Open proposals

Was this guidance useful?

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