ralph-operations
Use when managing Ralph orchestration loops, analyzing diagnostic data, debugging hat selection, investigating backpressure, or performing post-mortem analysis
- License 许可
- MIT
- 最近更新
- 2026-08-23
- Stars
- ★ 1
- Forks
- 0
01 前言 Front Matter
- name
- ralph-operations
- tags
- [ "loops", "diagnostics", "debugging", "analysis" ]
- description
- Use when managing Ralph orchestration loops, analyzing diagnostic data, debugging hat selection, investigating backpressure, or performing post-mortem analysis
02 资源文件 1
- 参考文档_meta.json286 B
03 SKILL.md 正文
Ralph Operations
Manage, monitor, and diagnose Ralph orchestration loops.
Loop Management
Quick Reference
| Task | Command |
|---|---|
| List active loops | ralph loops |
| List all (including merged) | ralph loops --all |
| View loop changes | ralph loops diff <id> |
| View loop logs | ralph loops logs <id> |
| Follow logs live | ralph loops logs <id> -f |
| Stop running loop | ralph loops stop <id> |
| Merge completed loop | ralph loops merge <id> |
| Retry failed merge | ralph loops retry <id> |
| Abandon loop | ralph loops discard <id> |
| Clean stale processes | ralph loops prune |
Loop ID format: Partial matching works - a3f2 matches loop-20250124-143052-a3f2
Loop Status
| Status | Color | Meaning |
|---|---|---|
| running | green | Loop is actively executing |
| queued | blue | Completed, waiting for merge |
| merging | yellow | Merge in progress |
| needs-review | red | Merge failed, requires intervention |
| merged | dim | Successfully merged (with --all) |
| discarded | dim | Abandoned (with --all) |
Starting & Stopping
Loops start automatically via ralph run:
- Primary loop: Runs in main workspace, holds
.ralph/loop.lock - Worktree loop: Created when primary is running, isolated in
.worktrees/<loop-id>/
ralph loops # Any loops running?
cat .ralph/loop.lock 2>/dev/null # Primary loop details
ralph loops stop <id> # Graceful stop
ralph loops stop <id> --force # Immediate stop
ralph loops discard <id> # Abandon + clean worktree
Inspecting Loops
ralph loops diff <id> # What changed
ralph loops logs <id> -f # Live event log
ralph loops history <id> # State changes
ralph loops attach <id> # Shell into worktree
Worktree context files (.worktrees/<loop-id>/):
| File | Contents |
|---|---|
.ralph/events.jsonl |
Event stream: hats, iterations, tool calls |
.ralph/agent/summary.md |
Current session summary |
.ralph/agent/handoff.md |
Handoff context for next iteration |
.ralph/agent/scratchpad.md |
Working notes |
.ralph/agent/tasks.jsonl |
Runtime task state |
Primary loop uses the same files at .ralph/agent/ in repo root.
Merge Queue
Flow: Queued → Merging → Merged or → NeedsReview → Merging (retry) or → Discarded
ralph loops merge <id> # Queue for merge
ralph loops process # Process pending merges now
ralph loops retry <id> # Retry failed merge
Reading state:
jq -r '.prompt' .ralph/loop.lock 2>/dev/null
tail -20 .ralph/merge-queue.jsonl | jq .
Diagnostics
Enabling
RALPH_DIAGNOSTICS=1 ralph run -p "your prompt"
Zero overhead when disabled. Output: .ralph/diagnostics/<YYYY-MM-DDTHH-MM-SS>/
Session Discovery
LATEST=$(ls -t .ralph/diagnostics/ | head -1)
SESSION=".ralph/diagnostics/$LATEST"
File Reference
| File | Contains | Key Fields |
|---|---|---|
agent-output.jsonl |
Agent text, tool calls, results | type, iteration, hat |
orchestration.jsonl |
Hat selection, events, backpressure | event.type, iteration, hat |
performance.jsonl |
Timing, latency, token counts | metric.type, iteration, hat |
errors.jsonl |
Parse errors, validation failures | error_type, message, context |
trace.jsonl |
All tracing logs with metadata | level, target, message |
Diagnostic Workflow
1. Errors first:
wc -l "$SESSION/errors.jsonl"
jq '.' "$SESSION/errors.jsonl"
jq -s 'group_by(.error_type) | map({type: .[0].error_type, count: length})' "$SESSION/errors.jsonl"
2. Orchestration flow:
jq '{iter: .iteration, hat: .hat, event: .event.type}' "$SESSION/orchestration.jsonl"
jq 'select(.event.type == "hat_selected") | {iter: .iteration, hat: .event.hat, reason: .event.reason}' "$SESSION/orchestration.jsonl"
jq 'select(.event.type == "backpressure_triggered") | {iter: .iteration, reason: .event.reason}' "$SESSION/orchestration.jsonl"
3. Agent activity:
jq 'select(.type == "tool_call") | {iter: .iteration, tool: .name}' "$SESSION/agent-output.jsonl"
jq -s '[.[] | select(.type == "tool_call")] | group_by(.iteration) | map({iter: .[0].iteration, tools: [.[].name]})' "$SESSION/agent-output.jsonl"
4. Performance:
jq 'select(.metric.type == "iteration_duration") | {iter: .iteration, ms: .metric.duration_ms}' "$SESSION/performance.jsonl"
jq -s '[.[] | select(.metric.type == "token_count")] | {total_in: (map(.metric.input) | add), total_out: (map(.metric.output) | add)}' "$SESSION/performance.jsonl"
5. Trace logs:
jq 'select(.level == "ERROR" or .level == "WARN")' "$SESSION/trace.jsonl"
Quick Health Check
SESSION=".ralph/diagnostics/$(ls -t .ralph/diagnostics/ | head -1)"
echo "=== Session: $SESSION ==="
echo -e "\n--- Errors ---"
wc -l < "$SESSION/errors.jsonl" 2>/dev/null || echo "0"
echo -e "\n--- Iterations ---"
jq -s 'map(select(.event.type == "iteration_started")) | length' "$SESSION/orchestration.jsonl"
echo -e "\n--- Hats Used ---"
jq -s '[.[] | select(.event.type == "hat_selected") | .event.hat] | unique' "$SESSION/orchestration.jsonl"
echo -e "\n--- Backpressure Count ---"
jq -s 'map(select(.event.type == "backpressure_triggered")) | length' "$SESSION/orchestration.jsonl"
echo -e "\n--- Termination ---"
jq 'select(.event.type == "loop_terminated")' "$SESSION/orchestration.jsonl"
Troubleshooting
Stale Processes
ralph loops shows loops that aren't running → ralph loops prune
Orphan Worktrees
.worktrees/ has directories not in ralph loops → ralph loops prune or git worktree remove .worktrees/<id> --force
Merge Conflicts
Loop stuck in needs-review:
ralph loops diff <id>— see conflicting changesralph loops attach <id>— resolve manually, commit, retryralph loops discard <id>— abandon if not worth fixing
Lock Stuck
"Loop already running" but nothing is → rm .ralph/loop.lock (safe if process is dead)
Agent Stuck in Loop
jq -s '[.[] | select(.type == "tool_call")] | group_by(.name) | map({tool: .[0].name, count: length}) | sort_by(-.count)' "$SESSION/agent-output.jsonl"
Red flag: Many iterations with few events = agent not making progress.
Merge Stuck in "merging"
Process died mid-merge. Unblock:
echo '{"ts":"'$(date -u +%Y-%m-%dT%H:%M:%S.000000Z)'","loop_id":"<loop-id>","event":{"type":"needs_review","reason":"Merge process died"}}' >> .ralph/merge-queue.jsonl
ralph loops discard <loop-id>
Worktree Corruption
git worktree repair
ralph loops prune
Cleanup
ralph clean --diagnostics # Delete all sessions
ralph clean --diagnostics --dry-run # Preview deletions
04 同仓库更多 Skills
crowd-prompting
A marketplace where AI agents improve prompts, system instructions, tool descriptions, and other text-based content with domain expertise from real-world operations — and earn tokens for valuable contributions.
dilemm-ai
Compete in DilemmAI, the prisoner's dilemma AI arena at dilemm.ai. Use when an OpenClaw agent wants to create an account, design and submit strategy prompts for their bot, enter matchmaking, analyze historical matches and leaderboards, and iterate on bot strategy. Requires headless browser automation (no API available yet).
arxiv-paper-reviews
Interact with arXiv Crawler API to fetch papers, read reviews, and submit comments. Use when working with arXiv papers, fetching paper lists by date/category/interest, viewing paper details with comments, or submitting paper reviews via API at http://150.158.152.82:8000.
arxiv-paper-reviews
Interact with arXiv Crawler API to fetch papers, read reviews, and submit comments. Use when working with arXiv papers, fetching paper lists by date/category/interest, viewing paper details with comments, or submitting paper reviews via API at http://150.158.152.82:8000.
git-summary
Get a quick summary of the current Git repository including status, recent commits, branches, and contributors.
agent-constitution
Interact with AgentConstitution governance contracts on Base Sepolia. Check compliance, read rules, log actions, query governance state.