mcpField
C
cost-optimize

cost-optimize

github.com/ruvnetAgent workflows80.0k745 installsUpdated today
Repository

About cost-optimize

cost-optimize is a skill for analyzing recent token usage across agents and models, then estimating savings from specific changes. It covers model fit, caching, redundant work, and reporting prioritized recommendations.

Key features

  1. 01Analyze recent token usage
  2. 02Assess model choice by task complexity
  3. 03Check cache hit rates
  4. 04Find overlapping agent work
  5. 05Estimate savings and prioritize changes
  6. 06Store patterns and emit model outcomes

Use cases

  1. 01Investigate unexpectedly high usage costs
  2. 02Find model downgrades for simple tasks
  3. 03Improve low prompt-cache hit rates
  4. 04Reduce duplicated agent work

MCP support

The skill uses MCP memory search and pattern-search tools to inspect usage history and prior optimization patterns, and memory or pattern-store tools to save recommendations.

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/ruvnet/ruflo --skill cost-optimize

After installing, type/cost-optimizein Claude Code to confirm it loaded.

Details

From the project README

Cost Optimize

Analyze recent token usage across agents and models, identify waste, and recommend specific optimizations with estimated dollar savings.

When to use

When costs are higher than expected or you want to proactively reduce spending. Analyzes model selection efficiency, cache utilization, agent redundancy, and prompt efficiency.

Steps

  1. Load usage data -- call mcp__plugin_ruflo-core_ruflo__memory_search on the cost-tracking namespace (last 7 days). The memory_* tools route by namespace; use them — not agentdb_hierarchical-* (which routes by tier).

  2. Analyze model fit -- for each agent, assess whether the model tier matches task complexity:

    • Agents doing simple tasks (formatting, linting) on Sonnet/Opus → suggest Haiku or Agent Booster
    • Agents doing complex tasks (architecture, security) on Haiku → flag quality risk

  3. Check cache rates -- compute cache hit rate per agent; if below 60%, recommend enabling or improving prompt caching (90% cost reduction on cache reads)

  4. Detect redundancy -- look for multiple agents performing overlapping tasks, or agents being spawned for work that could be batched

  5. Estimate savings -- for each recommendation, calculate: current cost, projected cost after optimization, dollar savings, percentage reduction

  6. Search prior optimization patterns -- call mcp__plugin_ruflo-core_ruflo__agentdb_pattern-search (ReasoningBank-routed; don't pass a namespace argument — pattern-* tools ignore it).

  7. Store the optimization pattern -- two paths:

    • Pattern store (typed, recommended): mcp__plugin_ruflo-core_ruflo__agentdb_pattern-store with type: 'cost-optimization'. Don't pass a namespace arg — ReasoningBank routes it; on bridge unavailability the fallback writes to the reserved pattern namespace with controller: 'memory-store-fallback' (see ruflo-agentdb ADR-0001).
    • Plain store (namespace-routable): mcp__plugin_ruflo-core_ruflo__memory_store --namespace cost-patterns — this DOES respect the cost-patterns namespace because memory_* is namespace-routed.

  8. Close the routing feedback loop — auto-emit hooks_model-outcome -- for each downgrade recommendation, format the outcome-emit command as part of the recommendation table so it can be run directly:

    # success path (downgrade worked)
node plugins/ruflo-cost-tracker/scripts/outcome.mjs "<task-description>" <model> success

# escalated path (had to upgrade after downgrade attempt)
node plugins/ruflo-cost-tracker/scripts/outcome.mjs "<task-description>" <model> escalated

    The script wraps npx @claude-flow/cli hooks model-outcome -t ... -m ... -o ... with explicit-argv spawnSync so quoting is safe. Without this signal the router does not learn from cost-tracker's recommendations and the booster bypass rate (see cost-booster-route skill) does not improve over time. This is the typed equivalent of the legacy routing-outcomes namespace (see ruflo-intelligence ADR-0001 §"Neutral").

  9. Report -- display: ranked recommendations with savings estimate, total potential savings, implementation priority (quick wins first), and any model-outcome events emitted in step 8

CLI alternative

npx @claude-flow/cli@latest memory search --query "cost optimization strategies" --namespace cost-patterns
npx @claude-flow/cli@latest memory store --key "opt-2026-05-04" --value '{...}' --namespace cost-patterns

Comments

Markdown supported