Prompt #143

Variables

wave_count, task_description, model_a, subtask_a, artifact_a, model_b, subtask_b, artifact_b, model_c

Tags

stack-aware,rag-stack,agent-orchestrate,ap50,multi-agent,waves

Source

research-2026-05-01-stack-aware

Use count

0

Created

2026-05-01T18:30:48.776731+00:00

Updated

2026-05-01T18:30:48.776731+00:00

Content

Plan and execute a {wave_count}-wave multi-agent task dispatch using bin/agent-orchestrate.

Context:
- agent-orchestrate implements AP50 multi-project dispatcher
- Tokens stored at /etc/gitea-tokens/<agent>.token (mode 0600), Base64-Basic format
- AP3 gateway (bin/agent-gateway) enforces: role+risk, data-class, secret-scan, tool-chain, cost circuit
- Per-session cost cap: $0.50; per-day: $5 cloud

Wave plan for task: {task_description}

Wave 1 (parallel, no dependencies):
- agent-{model_a}: {subtask_a} → outputs: {artifact_a}
- agent-{model_b}: {subtask_b} → outputs: {artifact_b}

Wave 2 (depends on Wave 1):
- agent-{model_c}: integrate {artifact_a} + {artifact_b}

For each agent: specify recommended_model, expected risk_level, escalation condition.
After dispatch: verify at https://orchestrator.joelduss.xyz/governance/agents