Default CLAUDE_TIMEOUT_MINUTES=15 too short for complex stories

[LarsCowe]

Problem

When Ralph starts a complex story that requires significant codebase exploration, the default 15-minute timeout (CLAUDE_TIMEOUT_MINUTES=15) kills the Claude Code process mid-execution (exit code 124).

This is especially problematic for:

• First loop on a new story (Claude needs to read many files to understand context)
• Stories with broad scope (e.g., "Timezone Detection & Display")
• Ultimate review mode where the review itself consumes part of the time budget

Reproduction

1. Start Ralph on a story that touches multiple files/modules
2. Claude begins exploring the codebase in loop [P0] mergeClaudeMd truncates user content after BMAD section #1
3. After 15 minutes, the timeout command kills Claude (exit 124)
4. Ralph logs show "stopped (exit 124)" with no progress

Suggestion

Increase default to 25-30 minutes

Modern Claude Code sessions with Opus regularly take 15-25 minutes on complex tasks. A 15-minute default causes premature termination on any non-trivial story.

Suggested default: CLAUDE_TIMEOUT_MINUTES=25

Consider adaptive timeouts

• First loop on a new story: Allow extra time (1.5x-2x) since context loading is heavier
• Subsequent loops with session continuity: Context is already loaded, so the default timeout is more appropriate
• Review loops: These are typically shorter — could use a separate REVIEW_TIMEOUT_MINUTES setting

Better feedback on timeout

When exit code 124 occurs, Ralph should:

1. Log clearly that a timeout happened (not just "stopped (exit 124)")
2. Not count a timeout as "no progress" for the circuit breaker — it's a config issue, not stagnation
3. Suggest increasing CLAUDE_TIMEOUT_MINUTES in the error message

Environment

• bmalph version: 2.10.0
• Platform: Windows 11 (bash via Git Bash)
• Driver: claude-code
• Model: Claude Opus 4.6