Saturday, November 7, 2026 270 picks · 133 builders credited Subscribe
AgentRags.

The agentic rag, wrung out.

ReleasesSaturday, October 3, 2026Trended 2 nights

OpenClaw v2026.9.8: replies land once; GPT-6.1 Sol does not

Official v2026.9.8 (43 PRs, 8 contributors named on the post; docs say the same scale). Delegated agent results return to the conversation that asked, once: inline if settled, or a single later delivery. Internal sessions no longer finish on NO_REPLY. Codex catalog processes hold fewer in-memory shell keys. Windows updates retry file locks; macOS npm updates accept aliased install paths. Docs are explicit that GPT-6.1 Sol was prepared (1.05M context, 128k output, 272k active input budget in OpenClaw) and was not complete in this release. Hannes Rudolph confirmed the miss: the bundled Codex CLI is 0.158.0 and needs a newer build. He posted three workarounds (updated ChatGPT app for computer use, force the OpenClaw runtime, or pin a 0.159.1+ Codex binary) and said 2026.9.9 is the fix. Not shipped by end of day.

Cover art for Releases: OpenClaw v2026.9.8: replies land once; GPT-6.1 Sol does not

Why it matters

Operators who upgraded for Sol will not get it from 9.8. The part that did ship is the multi-agent return path, which is the failure mode people actually hit.

From the source

OpenClawCodex CLI#openclaw#2026.9.8#gpt-6.1-sol#subagents#windows#codex

Related

More Releases

ReleasesOct 27

OpenClaw 2026.10.7 — Session Recovery Improvements

Latest stable (2026.10.7, ~Oct 24) focuses on replay-safe session recovery: interrupted sessions keep recovery provenance and unfinished tool outcomes, stopped conversations clean terminal residue, and various Codex/Claude CLI edge cases are tightened. Also safer update activation and database/worker continuity.

ReleasesOct 23

Hermes Agent v0.22.5 Ships Major Architecture Shifts

v0.22.5 (stable, Oct 23) moves hermes mcp serve, import-agent, OpenClaw migration, and Singularity backend out of core into official plugins. Adds per-task subagent routing (model or profile via advanced-delegation plugin), usage-window gauges for Nous plans (/usage bars), MCP tool cost declarations with per-chat spend limits, Edge TTS fix, Pillow 13 security bumps, and Electron 44 desktop updates. Detailed 10-item breakdown covers before/after behavior.

ReleasesNov 7

Real-World OpenClaw Usage and Ongoing Skill Security Considerations

Community reports of productive daily use (e.g., business workflows after sustained setup effort). Parallel ongoing discussion of skill marketplace risks—VirusTotal partnership for ClawHub scanning exists, yet independent audits continue to surface issues with malicious or high-risk skills. Local-first design remains the core value prop.

Never miss a night

Every pick, delivered to your reader.

One feed for everything, or one per agent harness you run. Rebuilt every night, with full credit on every item.