Radar brief — 2026-08-13
Agent runtimes tighten control planes this week: self-hosted Claude sessions, Codex plugin catalogs + safer auto-approve, and Cline’s human-gated plan mode—while Goose and Zed ship deeper loop and editor ergonomics.
What the evidence shows
Versioned GitHub releases dominate the high-signal set (Codex, Claude Code, Goose, Cline, Zed, Gemini CLI). One HTML changelog (Windsurf) and commit/nightly noise sit lower confidence.
claude self-hosted-runner turns own machines/containers into run targets for web/mobile/desktop (Team/Enterprise); archive plugin install over HTTPS with optional SHA-256 pin; Bedrock region-prefix env--approve-for-me auto-reviewed approvals; import Cursor skills / sync Claude & Cursor threadsgit.diff_base (HEAD vs default-branch merge base); drag-out from Project Panel (macOS/Linux Wayland); JSX/TSX linked editing + Emmet in arrow/return bodies@ path resolution; release-pipeline hardeningCross-cutting themes from tags: model / agent / mcp / cli lead; security and breaking cluster on Codex auto-approve defaults, Cline plan-mode blocks, Gemini path blocklist, and Claude self-hosted + pinned plugin archives.
For developers
- Codex CLI: Try plugin install from local/personal/workspace/remote catalogs on
rust-v0.147.0; use conversation sections and thread fork before long refactors. Treat--approve-for-meas explicit risk—pair withrust-v0.146.1cyber-model review defaults, not as blanket--full-auto. - Claude Code: If on Team/Enterprise, pilot
claude self-hosted-runnerfor private network tool access; preferarchiveplugins with SHA-256 pins over ad-hoc git/npm sources. - Cline: Upgrade to
v4.1.4before relying on plan mode—you drive plan→act, and mutating shell/file ops are hard-blocked. On Desktopv0.0.8, use edit-earlier-message + checkpoint rewind instead of manual git reset when a turn goes sideways. - Goose
v1.46.0: Use streaming shell + per-message cost/TTFT stats to catch runaway loops early; cache-safe turn assembly matters if you depend on prompt caching. - Zed AI
v1.15.0: Setgit.diff_basedeliberately (uncommitted vs branch-merge-base) so agent/review gutters match your workflow; JSX/TSX linked editing reduces rename churn. - Gemini CLI
v0.55.1: Pull for sensitive-path blocklist + path-resolution fixes; skipv0.56.0-preview/nightlyunless you are testing caretaker internals. - OpenCode
v1.18.10: Useful if you need Modal model discovery and stabler tab/session chrome—smaller functional delta than the agent-runtime releases above.
For teams
- Enterprise auth & residency: Antigravity
1.1.10Business sign-in (Gemini Enterprise seats, region, admin controls) and Claude Code self-hosted runners are the concrete fleet levers this window—map them to your IdP and data-boundary policies before widening seat count. - Approval policy: Codex’s
--approve-for-meplus cyber-model auto-review defaults need a written allowlist (repos, network, secrets). Default-deny on new plugin marketplaces (Bedrock/Claude Code/workspace publish fromrust-v0.146.0). - Plan-mode standard: Cline’s hard blocks in
v4.1.4are a good template for “investigate ≠ edit” in shared coding agents—encode the same expectation in playbooks even if you standardize on another assistant. - Supply chain: Prefer Claude
archive+ SHA-256 pinning and Codex portable plugins with catalog review over unmanaged skill drops; inventory Cursor→Codex skill imports so ownership is clear. - Cost observability: Goose’s per-message usage/cost UI is a practical pilot metric for chargeback or runaway-session alerts.
- Desktop fleets: Windsurf’s restart confirmation when local agents are busy reduces blown in-flight work; still validate from primary GitHub sources before mandating IDE changelog items.
- Open proposals (P5 backlog): older Codex
rust-v0.145.0/v0.144.6, Clinev4.0.7, Goosev1.44.0, OpenCodev1.18.5, Zed pre, Grok sync—triage only after the newer versions in this brief.
What to ignore
- Gemini CLI
v0.56.0-preview.1and nightlies — caretaker/Firestore/Pub/Sub scaffolding; not a stable developer interface signal. - xAI Grok
be71313commit — monorepo sync without a clean semver release story; UX/presence bits only until tagged. - Windsurf HTML changelog as sole proof — fine for product UX notes, weaker than GitHub release tags in this mix (~55% GitHub/changelog trust).
- OpenCode toast/tab polish relative to plugin, sandbox, and plan-mode security changes elsewhere.
- Inflated “+N more changes” tails without named behaviors—don’t plan migrations from count noise.
- Stale open proposals pointing at versions superseded by this brief’s Codex
0.146.x/0.147.0, Clinev4.1.4, Goosev1.46.0.
Watch next
- Whether Codex plugin manifests +
--approve-for-mepick up tighter org policy hooks after the cyber-model review backport. - Adoption path for Claude Code self-hosted runners (permissions, sandbox, compaction tags already on the release) beyond Team/Enterprise docs.
- Cline Desktop checkpoint fork reliability under multi-root workspaces and long compaction chains.
- Goose unrolled loop + cache semantics under high-MCP tool chatter (tags already mark MCP/skills/hooks).
- Gemini CLI’s tool-registry discovery landing cleanly in a non-preview
v0.56.xwith the path-blocklist fixes retained. - Zed follow-ons that connect
git.diff_basemore directly to agent review flows (MCP/agent tags present, feature text still editor-first). - Antigravity Enterprise admin-control surface area once Business sign-in is in wider use.
How to use DevAgentRadar
- Read What the evidence shows as the only source of truth for versions—click through to GitHub tags before changing defaults.
- Developers apply one assistant upgrade at a time and verify plan/approve/sandbox behavior on a throwaway repo.
- Teams convert security-tagged items (auto-approve, plan hard-blocks, path blocklists, pinned plugin archives, self-hosted runners) into policy checklists, not optional tips.
- Ignore prerelease/nightly/commit-only rows unless you are reproducing a specific bug.
- Feed outcomes back via open improvement proposals; prioritize against this week’s newer tags, not superseded P5 links.
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
-
Zed AI
v1.15.0: This week's release includes a newgit.diff_basesetting for choosing whether editor gutters, file-status colors, andgit: diffshow uncommitted changes aga — source -
Claude Code
v2.1.224: Added self-hosted environments:claude self-hosted-runnerturns 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 -
Goose
v1.46.0: Unrolled agent loop #9574 · Cache-safe request assembly with append-only turn context and declared cache semant — source -
OpenAI Codex CLI
rust-v0.146.0: Name new sessions with/newor/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 -
Gemini CLI
v0.55.1: fix/verify release npm ci ignore scripts · fix(ci): prevent workspace binary shadowing in release verification · Feat/tool registry discovery · fix(ci): prevent — 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
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— 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 -
Gemini CLI
v0.56.0-preview.1: Changelog for v0.55.0-preview.1 · chore(release): bump version to 0.56.0-nightly.20260806.g761f604c1 · Changelog for v0.54.0 · Reclassifying Capacity Exhaustion — 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
v0.56.0-nightly.20260808.gcf22ac7e8: Reclassifying Capacity Exhaustion as Terminal Error · feat(caretaker): update Firestore schema with error, and pr_number fields · feat(caretaker-triage): prompt — source -
xAI Grok CLI (Grok Build)
be71313— Synced from monorepo: Synced from monorepo · Presence protocol end-to-end: live presence updates through the gateway to client presence tiers · Worktree: narrow standalone origin fet — source
Open proposals
-
Improve with OpenAI Codex CLI rust-v0.145.0: rust-v0.145.0 0.145.0 (audience: team, P5)
-
Improve with Cline v4.0.7: v4.0.7 (audience: team, P5)
-
Improve with xAI Grok CLI (Grok Build) 8adf901: Synced from monorepo (audience: team, P5)
-
Improve with OpenCode pr-38252-videos: pr-38252-videos PR #38252 verification videos (audience: everyone, P5)
-
Improve with OpenAI Codex CLI rust-v0.144.6: rust-v0.144.6 0.144.6 (audience: developer, P5)
-
Review breaking changes in Zed AI v1.13.0-pre (audience: team, P5)
-
Improve with OpenCode v1.18.5: v1.18.5 (audience: team, P5)
-
Improve with Goose v1.44.0: v1.44.0 (audience: team, P5)
Anonymous · one vote per visitor · re-click to clear. Helps improve digests—not a product ranking.