Anthropic’s Own Staff Can’t Use Claude Code’s ‘Bypass Permissions’ Mode, Client Code Shows
Anthropic's public JavaScript separates unapproved-control refusals from outdated-client errors; another handler says employee sessions requesting bypass mode start in Accept edits.
By Ryan Merket · Published
RUNTIMEWIRE INVESTIGATION — Original analysis
Original reporting by RuntimeWire, based on reverse engineering, testing.
Why it matters
The code exposes specific employee-facing error handling, while leaving the server rules and customer access to equivalent controls unresolved.
Reporting record
Finding
RuntimeWire's inspection found that Claude Code's public client maps staff-linked unapproved-controls refusals separately from outdated-client errors, and that a separate session handler reports employee sessions requesting bypass mode as starting in Accept edits.
How we verified
Methods: reverse engineering, testing.
The inspected public JavaScript assets include the API-error parser, staff-facing error messages, and a separate session-start handler; RuntimeWire also preserved the retrieved assets' bytes and SHA-256 hashes.
RuntimeWire inspected the publicly distributed Claude Code JavaScript assets, traced the parser's error categories to their staff-facing messages, examined the separate employee-account session-start branch, and preserved the retrieved files' bytes and SHA-256 hashes. No live API response was observed.
Reproduction
RuntimeWire was not able to independently reproduce the finding.
Anthropic's publicly distributed Claude Code client contains separate paths for refusing staff sessions with unapproved security controls and for identifying outdated clients, according to RuntimeWire's October 18th inspection. The code shows how the client classifies and explains these conditions.…
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.