[randomparity]

report utilization now names only the supplied global connection options in its refusal, using the existing canonical option identities. The CLI guide lists the complete refused set, aliases and boolean polarities, and distinguishes the report-local profile selector.

Public CLI regression tests cover the eight identities, short aliases, negative switches, repeated flags and mixed options. They assert exact names without argument values, exit 2, and no profile inventory, loading, survey or report file. Restoring the old fixed list causes all 14 cases to fail. The report, HTML and CLI help checks passed 194 tests.

The diagnostic retains canonical positive long names rather than original alias spelling, and earlier validation errors still take precedence. Configuration and acceptance behavior are unchanged.

Validation: just verify passed once (1150.418 seconds, including the configured coverage gate, MCP smoke and artifact verification); pinned uv run --no-sync prek run --all-files passed (107.737 seconds).

Requested by #1892 under campaign 152593d7a535-fdfc969e-824d-4d69-a9ae-3ce95496903a.

Closes #1892

Review exit payloads

• Accepted failure-model limit: existing option identities produce canonical positive long names, including for aliases and negative boolean switches.
• Accepted failure-model limit: earlier validation and exported-setting refusals retain precedence over the global-option diagnostic.
• No open findings, deferred work or follow-up candidates.

[randomparity]

verdict: approve
exit: none
findings: 0
iterations: 1
security: not triggered
delivered-head-sha: 761461c

Forge whole-branch review

forge review: not required (Cast direct execution leaves whole-branch review to quest)

Review exit payloads

• Accepted failure-model limit: existing option identities produce canonical positive long names, including for aliases and negative boolean switches.
• Accepted failure-model limit: earlier validation and exported-setting refusals retain precedence over the global-option diagnostic.
• No open findings, deferred work or follow-up candidates.