Menu
Read accumulated cost-tracking spend + budget config, compute utilization, emit 50/75/90/100% alert ladder
| name | cost-budget-check |
| description | Read accumulated cost-tracking spend + budget config, compute utilization, emit 50/75/90/100% alert ladder |
| argument-hint | [--period today|week|month|all] |
| allowed-tools | Bash mcp__claude-flow__memory_retrieve mcp__claude-flow__memory_list mcp__claude-flow__memory_store |
Reads cost-tracking:budget-config for the project's budget limit, sums total_cost_usd across session-* records produced by cost-track, computes utilization, and emits the measured 4-tier alert ladder (50% INFO / 75% WARNING / 90% CRITICAL / 100% HARD_STOP).
Until P2 (this skill) landed, the README documented the alert ladder but no code checked it. Now this skill is the gate.
/cost-optimize first./loop 30m for continuous monitoring.Run the check:
node plugins/ruflo-cost-tracker/scripts/budget.mjs check
Filter by period: BUDGET_PERIOD=today (default all). Use BUDGET_QUIET=1 for machine-readable JSON.
Inspect the markdown summary — budget, spent, remaining, utilization percentage, alert level (🟢 OK · 🟡 INFO · 🟠 WARNING · 🔴 CRITICAL · 🛑 HARD_STOP), and the recommended action.
Set / inspect the budget:
node plugins/ruflo-cost-tracker/scripts/budget.mjs set 50.00
node plugins/ruflo-cost-tracker/scripts/budget.mjs get
HARD_STOP path — budget.mjs check exits with code 1 when utilization ≥ 100%. Wrap critical agent spawns in a budget.mjs check && spawn … guard to fail closed.
cost-tracking:budget-config){
"budget_usd": 50.00,
"setAt": "2026-05-05T...",
"thresholds": { "info": 0.50, "warning": 0.75, "critical": 0.90, "hard_stop": 1.00 }
}
| Threshold | Level | Action |
|---|---|---|
| 50% | INFO 🟡 | log notification, no UX disruption |
| 75% | WARNING 🟠 | display warning, suggest /cost-optimize |
| 90% | CRITICAL 🔴 | urgent alert, recommend model downgrades |
| 100% | HARD_STOP 🛑 | halt non-essential spawns; exit code 1 |
cost-track (producer) — populates cost-tracking:session-*cost-report — same data source, narrative formatcost-optimize — recommended action when WARNING/CRITICALSpawn nested sub-agents (agents that spawn sub-agents, up to depth=5) via Claude Code's native Task tool — for context-managed deep delegation
Author a workflow — either an MCP workflow template (persisted, lifecycle) or a native .claude/workflows/*.js orchestration script (agent/parallel/pipeline fan-out)
Run a workflow — drive an MCP workflow lifecycle (execute/pause/resume/cancel) or invoke + resume a native .claude/workflows/*.js orchestration via the Workflow tool
Side-by-side comparison of ruflo vs HAL vs other GAIA harnesses — capability gaps, design decisions, and improvement roadmap
Diagnose why a GAIA question failed — extract trace, classify failure mode, and propose a fix
Walk through a complete GAIA benchmark→submit flow — from key resolution through HAL-compatible package generation