agents

name	agents
description	Agent swarm development dashboard — live Railway containers, issue queue, PR pipeline, JSONL events, pool utilization, and queue drain status. Use when checking agent dev tasks, spawning status, or monitoring issues #315-#319.

🤖 Agent Swarm Observatory

📡 Railway Service Pools (Live)

!curl -s -X POST "https://railway.com/graphql/v2" -H "Authorization: Bearer $(grep RAILWAY_API_TOKEN /Users/playra/trinity-w1/.env | cut -d= -f2)" -H "Content-Type: application/json" -d '{"query":"query($id:String!){project(id:$id){services{edges{node{id name deployments(first:1){edges{node{status createdAt}}}}}}}}","variables":{"id":"aa0efa7f-95e6-4466-8de6-43945a031365"}}' 2>/dev/null | python3 -c " import sys,json,datetime d=json.load(sys.stdin) nodes=[e['node'] for e in d['data']['project']['services']['edges']] total=len(nodes) agents=[n for n in nodes if n['name'].startswith('agent-')] pool0='acfee27a-74e8-4436-961c-698ae93508ca' pool1='12c2bdf9-d124-4a45-93ad-22921e842d1b' p0=[n for n in nodes if n['id']==pool0] p1=[n for n in nodes if n['id']==pool1] def st(n): if n['deployments']['edges']: return n['deployments']['edges'][0]['node']['status'] return 'NO_DEPLOY' print(json.dumps({ 'total_services': total, 'pool_0_ubuntu': {'id': pool0[:8], 'status': st(p0[0]) if p0 else 'NOT_FOUND'}, 'pool_1_agents_anywhere': {'id': pool1[:8], 'status': st(p1[0]) if p1 else 'NOT_FOUND'}, 'agent_services': [{'name':n['name'],'status':st(n)} for n in agents], 'active_building': len([n for n in nodes if st(n) in ('DEPLOYING','BUILDING')]), 'slots_used': len(agents), 'slots_free': 10-len(agents) }, indent=2)) " 2>/dev/null || echo "⚠️ Railway API unavailable"

📋 Agent Issues (agent:spawn + agent:queued)

!echo "=== SPAWNING ==="; gh issue list --repo gHashTag/trinity --label "agent:spawn" --state open --limit 20 --json number,title,labels,assignees,createdAt --jq '.[] | "#\(.number) \(.title) [\(.labels | map(.name) | join(","))] \(.createdAt[:10])"' 2>&1 || echo "gh unavailable"; echo ""; echo "=== QUEUED ==="; gh issue list --repo gHashTag/trinity --label "agent:queued" --state open --limit 20 --json number,title,createdAt --jq '.[] | "#\(.number) \(.title) \(.createdAt[:10])"' 2>&1 || echo "none"

🔀 Agent PRs (feat/issue- branches)

!gh pr list --repo gHashTag/trinity --state open --limit 15 --json number,title,headRefName,statusCheckRollup --jq '.[] | select(.headRefName | startswith("feat/issue-")) | "#\(.number) [\(.headRefName)] \(.title) checks:\(.statusCheckRollup | if . then (. | map(.conclusion // .status) | join(",")) else "none" end)"' 2>&1 || echo "No agent PRs"

📡 Recent Events (last 25 lines)

!if [ -f /Users/playra/trinity-w1/.trinity/cloud_events.jsonl ]; then tail -25 /Users/playra/trinity-w1/.trinity/cloud_events.jsonl | python3 -c " import sys,json for line in sys.stdin: line=line.strip() if not line: continue try: e=json.loads(line) ts=e.get('timestamp','?')[:19] typ=e.get('type','?') iss=e.get('issue','?') msg=e.get('message',e.get('event','')) emoji={'spawn':'🚀','kill':'💀','heartbeat':'💓','error':'❌','pr':'🔀','build':'🔨','test':'🧪','complete':'✅','queued':'⏳'}.get(typ,'📌') print(f'{ts} {emoji} #{iss} [{typ}] {msg}') except: print(line) " 2>/dev/null; else echo "No cloud events yet"; fi

🔄 GitHub Actions (agent workflows)

!gh run list --repo gHashTag/trinity --workflow agent-spawn.yml --limit 5 --json databaseId,status,conclusion,createdAt,headBranch --jq '.[] | "\(.createdAt[:16]) \(if .conclusion == "success" then "✅" elif .conclusion == "failure" then "❌" elif .status == "in_progress" then "🔄" else "⏳" end) \(.status)/\(.conclusion // "—") \(.headBranch // "—")"' 2>&1 || echo "No spawn runs"; echo "---"; gh run list --repo gHashTag/trinity --workflow agent-queue-drain.yml --limit 3 --json status,conclusion,createdAt --jq '.[] | "\(.createdAt[:16]) \(if .conclusion == "success" then "✅" else "⏳" end) drain: \(.status)/\(.conclusion // "—")"' 2>&1 || echo "No drain runs"

🏗️ Container Image

!gh api user/packages/container/trinity-agent/versions --jq '.[0] | "📦 trinity-agent:latest — updated \(.updated_at[:10]) tags: \(.metadata.container.tags | join(","))"' 2>/dev/null || echo "⚠️ GHCR package not accessible"

📊 Local Agent State

!if [ -f /Users/playra/trinity-w1/.trinity/cloud_agents.json ]; then cat /Users/playra/trinity-w1/.trinity/cloud_agents.json | python3 -c " import sys,json,datetime d=json.load(sys.stdin) agents=d if isinstance(d,list) else d.get('agents',[]) active=[a for a in agents if a.get('active',False)] inactive=[a for a in agents if not a.get('active',False)] print(f'Active: {len(active)}/{len(agents)}') for a in active: iss=a.get('issue','?') sid=a.get('service_id','?')[:8] ts=a.get('created_at','?') print(f' 🟢 #{iss} svc:{sid} started:{ts}') for a in inactive[-3:]: iss=a.get('issue','?') print(f' ⚪ #{iss} (done)') " 2>/dev/null; else echo "No local state file"; fi

🎯 Issue → Pool Mapping (round-robin)

!echo "Pool 0 (ubuntu/acfee27a): even issues"; echo "Pool 1 (Agents Anywhere/12c2bdf9): odd issues"; echo "---"; gh issue list --repo gHashTag/trinity --label "agent:spawn" --state open --limit 20 --json number,title --jq '.[] | "#\(.number) → Pool \(.number % 2) \(if .number % 2 == 0 then "(ubuntu)" else "(Agents Anywhere)" end) — \(.title[:60])"' 2>&1 || echo "gh unavailable"

Task

Analyze the data above and present a rich Agent Swarm dashboard with emojis.

Focus area: $ARGUMENTS (default: status)

Dashboard Format

ALWAYS output the full dashboard — never compress to one line. Use this format:

🤖 ═══════════════════════════════════════════════════
   TRINITY AGENT SWARM — DEVELOPMENT OBSERVATORY
   ═══════════════════════════════════════════════════

📡 SERVICE POOLS
   Pool 0 (ubuntu):           [status emoji] [deployment status]
   Pool 1 (Agents Anywhere):  [status emoji] [deployment status]
   Active builds: N | Free slots: N/10

📋 ISSUE QUEUE
   🚀 Spawning:
     #N — title — pool — status
   ⏳ Queued:
     #N — title — waiting since [date]

   Queue depth: N spawning + N queued = N total

🔀 AGENT PRs
   #N — [branch] — title — checks: [status]
   ...
   Open: N | Merged today: N

📡 LIVE EVENTS (last 10)
   [timestamp] [emoji] #issue [type] message
   ...

🔄 GITHUB ACTIONS
   agent-spawn: [last 5 runs with status]
   queue-drain: [last 3 runs with status]

🗺️ POOL MAPPING (round-robin: issue# % 2)
   Pool 0 (even): #316, #318, ...
   Pool 1 (odd):  #315, #317, #319, ...
   ⚠️ Contention: N issues compete for Pool X

📦 INFRASTRUCTURE
   GHCR image: [status + last updated]
   Local state: N active / N total agents
   Events log: N entries

🎯 RECOMMENDATIONS
   [dynamic recommendations based on current state — see rules below]

⏱️ ESTIMATED PIPELINE
   [time estimate for queue drain based on 2 pools × ~1h per agent]

Recommendation Rules

If both pools DEPLOYING/BUILDING: → "⚠️ Both pools busy. Queue drain runs every 5min. ETA for next slot: ~Xmin" → Show which issues are running vs queued

If 1 pool free: → "🟢 Pool N free — next queued issue #M will auto-spawn via drain" → "💡 Manual: tri cloud spawn <N> to skip queue"

If both pools free + queued issues: → "🟢🟢 Both pools idle! Queue drain should trigger in ≤5min" → "💡 Manual: add agent:spawn label or run tri cloud spawn <N>"

If both pools free + no queued issues: → "✅ All clear. Ready for new agent tasks." → Suggest: check gh issue list --label agent:spawn

Contention warnings: → If >2 issues map to same pool: "⚠️ Pool X has N issues queued — sequential processing ~Nh" → Suggest rebalancing or manual spawn on other pool

Queue ETA Calculation

• Each agent takes ~1h (AGENT_TIMEOUT=3600s)
• 2 pools = 2 parallel agents
• Queue ETA = ceil(queued_issues / 2) × 1h
• Show: "⏱️ 5 issues ÷ 2 pools = ~3h total (2+2+1)"

Issue Status Tracking

For each tracked issue, show:

• 🔵 SPAWNING — container starting
• 🟢 RUNNING — claude code active
• 🟡 SELF-REVIEW — build/test/format
• 🔀 PR CREATED — waiting for review
• ✅ MERGED — done
• ❌ FAILED — needs attention
• ⏳ QUEUED — waiting for pool slot

If focus=spawn

Trigger: tri cloud spawn <N> and show result.

If focus=events

Show full event log (last 50 lines) with parsed JSONL.

If focus=pools

Deep dive into Railway API: service details, deployment history, env vars check.