What
A self-hosted runner is a machine you run yourself that starts Claude Code sessions as separate processes. When one of those processes fails, the runner now gives a clearer reason:
- A process that never started is reported as "child failed to start (): ..." instead of being treated as an exit.
- A start error of class
e2big, meaning the launch configuration was too large, counts as a session configuration error (SESSION_FAILURE_KIND_SESSION_CONFIG_ERROR) instead of a runner crash. - Messages for other start errors include the size of the longest argument and the longest environment entry.
- If the process dies before it finishes starting and its error output contains "disabled by your organization's managed settings (disableSideloadFlags)", the user sees "This machine's managed settings (disableSideloadFlags) refused the launch."
The disableSideloadFlags error text is now built from a shared constant. The wording is unchanged.
Why
Before, every failure was reported as the process exiting and classed as a runner crash. Operators can now tell a configuration problem, such as a launch that is too large or one blocked by the managed setting disableSideloadFlags, from a real crash.
The entry above is what we published on the day. These lines were added later, as Anthropic's own pages caught up, and they sit beside the original rather than replacing it.
Local and cloud sessions also load the skills enabled for your claude.ai account, except local sessions when your organization sets [`disableSideloadFlags`](/docs/en/settings-reference#disablesideloadflags). Cloud sessions load them instea…desktop see the edit
| [`disableSideloadFlags`](/docs/en/settings-reference#disablesideloadflags) | Reject the `--plugin-dir`, `--plugin-url`, `--agents`, and `--mcp-config` flags at startup. In cloud sessions, Claude Code instead starts the session and drops …managed-settings see the edit
A small documentation edit on Deploy managed settings touched a line naming disableSideloadFlags after this was published.