fix(cliagent): redact command error messages

[XnLemon]

Objective

Redact model-visible Codex and ClaudeCode CLI command error messages for Phase2 redaction closure.

Completed Changes

• Route Codex CLI run-error and flow-error ResponseError.Message through the shared platform redactor.
• Route Codex CLI error assistant content through the same redactor so command stdout/stderr secrets are not exposed to the model-visible stream.
• Route ClaudeCode CLI run-error and flow-error ResponseError.Message through the shared platform redactor.
• Route ClaudeCode CLI error assistant content through the same redactor.
• Preserve raw output hook semantics by keeping hook inputs as original stdout/stderr/error before model-visible redaction.
• Add regression coverage for Authorization, api_key, and Cookie leakage across run-error and flow-error paths, plus assertions that non-sensitive diagnostic context remains visible.

Validation

• go test ./agent/codex -run TestCodexAgent_Run_ResumeAndCreateErrorsReturnRunError|TestCodexAgent_Run_RawOutputHookReceivesCommandError|TestCodexAgent_Run_RawOutputHookError
• go test ./agent/claudecode -run TestClaudeCodeAgent_Run_CommandError|TestClaudeCodeAgent_Run_RawOutputHookError
• go vet ./agent/codex ./agent/claudecode
• git diff --check
• Subagent open-code-review first pass: 0 High / 0 Medium / 2 Low
• Subagent open-code-review final pass: 0 High / 0 Medium / 0 Low

Known Risks / Limitations

• Full go test ./agent/codex ./agent/claudecode is blocked in this Windows environment by the pre-existing TestExecCommandRunner_Run dependency on sh in PATH.
• This increment is scoped to Codex and ClaudeCode CLI command error events.

Follow-up Work

• Continue Phase2 closure by scanning remaining model-visible server, model-provider, and graph error surfaces.

[coderabbitai]

Warning

Review limit reached

@XnLemon, you've reached your PR review limit, so we couldn't start this review.

Next review available in: 21 minutes

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

How can I continue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.

How do review limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.

Please refer docs for additional details.

Review details

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 548af6a5-ed82-47ff-ac66-fea3fd5e86b0

📥 Commits

Reviewing files that changed from the base of the PR and between 361e2a4 and 4ceedac.

📒 Files selected for processing (4)

• agent/claudecode/claude_agent.go
• agent/claudecode/claude_agent_test.go
• agent/codex/codex_agent.go
• agent/codex/codex_agent_test.go

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch codex/phase2-codex-claude-error-redaction

---

_{Comment @coderabbitai help to get the list of available commands.}