fix(auth): ignore disabled Claude provider settings
Disabled Claude provider settings such as CLAUDE_CODE_USE_BEDROCK=false were counted as configured sign-in sources. Apply the same truthiness rules used for process environment switches to settings values, so disabled Bedrock, Vertex, and Foundry switches no longer create false sign-in evidence. Token/key presence detection remains intact. This detects configured sources, not live provider acceptance.
Regression covers disabled and enabled boolean, numeric, and string values for all three providers, plus preservation of an independent API key source.
Validation: the unfixed regression commit 312bf0aee611f2510dfd1ce89ede58b4b097e671 failed at CLAUDE_CODE_USE_BEDROCK=false in remote run 40618109178. The same command passed at final head f204394ba5a99fc1128648c313a0a5da8548050c in remote run 40618464774: one regression covering 33 provider/value combinations and API-key preservation. Command: cd cmux-tui && cargo test -p acpmux --test claude_auth_settings --locked through cmux-ci test. Independent Codex review passed on that exact head with no findings. Primary diff inspection, git diff --check, and scoped rustfmt --check passed.
Risk: limited to credential-source reporting and diagnostics. No credential writes, provider requests, or sign-in lifecycle changes.
Requested during the onboarding audit:
Also, we need to be fixing authentication issues, like the whole onboarding situation. Is that in? What does you improve? What does you fix? The edge cases, like you guys need to be comprehensively brainstorming all the onboarding edge cases: foot guns, paper cuts, snags, clips, whatever the fuck.
Changelog
Fixed: Disabled Claude provider settings no longer count as sign-in evidence.
Agent session: agent session