OpenClaw 2026.8.1 Adds Masked Prompts That Keep Credentials Out of Chat

OpenClaw 2026.8.1, released on August 31, 2026, lets an agent request a credential without its value entering chat or model context, and adds automation permissions granted once per exact operation.

OpenClaw AI-agents automation security coding

OpenClaw 2026.8.1, a release of the personal AI assistant of the same name, shipped on August 31, 2026¹. The GitHub repository carries more than 380,000 stars (checked September 1, 2026)².

An agent can now request a credential through a masked prompt, without the value entering chat or model context. An opt-in proxy limits protected-secret substitution to approved destinations¹
Recurring automation can be approved once for an exact operation, with that permission inspectable and revocable later, and a fresh approval required when the job or operation changes¹
Sessions can run on paired devices or cloud workers, moving the session workspace with them, and warm machines and project seeds can be reused for later cloud sessions¹

There are two breaking changes. The bundled OpenProse plugin and the /prose command are removed, and codex/* and openai-codex/* model refs, stored sessions, and automation routes migrate to openai/*. The OpenAI route migration is done with openclaw doctor --fix. For OpenProse, openclaw doctor --fix cleans stale configuration, and the upstream Agent Skill migration has to be followed as well; existing .prose source files are kept¹. For anyone wiring credentials into an agent workflow, having a path that keeps the value out of model context changes the available design choices.

Sources

OpenClaw 2026.8.1 - OpenClaw official release notes (August 31, 2026)
openclaw/openclaw - OpenClaw official repository (checked September 1, 2026)