Who contributed to OpenClaw on August 1, 2026?

4 developers shipped this update, including steipete, vincentkoc, openclaw-docs-sync[bot], and morluto.

What were the notable OpenClaw updates?

improve(update): reuse dev preflight build cache, fix(google): preserve canonical replay and provider error ownership, and fix(talk): bound durable transcript persistence.

Aug 1, 2026

$ the-wire · showcase

Dev preflight and final checkout share build cache; Google signing and Talk transcripts hardened

By RepoJournal · Filed 19:18 UTC on September 7, 2026 · About OpenClaw

4 people shipped this

steipete @steipete 2 cited

vincentkoc @vincentkoc 2 cited

morluto @morluto 1 cited

openclaw-docs-sync[bot] @openclaw-docs-sync[bot]

OpenClaw's dev-channel updater now reuses the build cache across preflight and final checkout, cutting duplicate TypeScript declaration and Control UI builds.

The dev-channel updater used to build the same TypeScript declarations twice on source-checkout updates, once in the safety preflight worktree and again after the live checkout moved, and it also rebuilt the Control UI after the full build had produced it. The dev preflight and final checkout now share build-all's content-addressed cache, so operators on the dev channel skip the duplicate work [1].

Google/Vertex replay handling now preserves the exact provider Part owning each thought signature across streamed text, thinking, tools, standalone signature carriers, and next-turn replay, keeping the mandatory empty `text: ""` discriminator on signed empty Parts. Malformed, array-shaped, and JSON-serialized historical function-call arguments are normalized through the shared transport coercion owner, restoring "actionable SDK failures" [2].

Durable Talk transcript persistence no longer retains an unbounded backlog when the Gateway persistence path stalls. A bounded serial queue at the Talk owner boundary keeps accepted writes FIFO, makes overflow terminal until explicit close, and adds capacity-waiting coordination in the durable client voice session [3]. Related work in Crabbox hardened provider reconciliation: quarantine exact coordinator-owned AWS and Azure orphan candidates across consecutive inventories before deletion, keep provider-tag-only and `keep=true` resources report-only, and add per-provider circuit backoff after inventory failures [4]. Also, RunPod leases with a configured `ssh.key` now get the matching `.pub` key serialized before pod creation, fixing the case where the pod reported `PUBLIC_KEY` as `null` and the SSH readiness check could not authenticate [5].

Action items

→ Update dev-channel source-checkout operators to pick up the shared preflight build cache; watch for faster update cycles. openclaw/openclaw [plan]
→ Review Google/Vertex streamed signing flows to confirm the preserved canonical replay and provider error ownership land as documented. openclaw/openclaw [plan]
→ If you run broker AWS/Azure leases or RunPod provisioning in Crabbox, verify image promotion and SSH key behavior after upgrade. openclaw/crabbox [plan]

References

[1] improve(update): reuse dev preflight build cache ↗ openclaw/openclaw
[2] fix(google): preserve canonical replay and provider error ownership ↗ openclaw/openclaw
[3] fix(talk): bound durable transcript persistence ↗ openclaw/openclaw
[4] fix(worker): harden provider reconciliation ↗ openclaw/crabbox
[5] fix(runpod): make new pods accept the configured SSH key ↗ openclaw/crabbox