$ loading_
Analyze worst-case token budgets for agent workflows and issue signed certificates.
Copy the install command and let the AI configure it · recommended for beginners
Please install the "costwright — agent budget analysis" MCP server from askskill: Run: claude mcp add 'io-github-hernaninverso-costwright-mcp' -- npx -y costwright-mcp
Run a worst-case token budget analysis for this LLM agent workflow: classify the user request, retrieve 5 documents, generate an answer, then perform one self-check. List per-step budget, total budget, and key risk points.
A step-by-step worst-case token breakdown, total budget estimate, and notes on stages that may exceed the limit.
Generate a signed budget certificate for this agent configuration: a GPT-style model, 3 tool-call rounds, a 2-retry policy, and a final summarization step. Output an auditable budget certificate and summary.
An auditable signed or verifiable budget certificate with a configuration summary and budget conclusion.
Compare worst-case budgets for two agent designs: Plan A uses one retrieval step with long-context generation, while Plan B uses multi-round retrieval with short-context iterations. Provide budget differences, risk ratings, and optimization suggestions.
A comparison of both plans' budgets, risk ratings, and recommendations to reduce token usage.
Discover live models and pricing to route tasks to compatible low-cost LLMs.
Estimate LLM token spend for coding tasks with ranges, scenarios, and confidence.
Perform cross-chain crypto actions, contract decoding, portfolio checks, and security screening.
Review contract code with multi-agent AI and get verdict-driven quality insights.
Explore ERC-8004 agent coverage, rankings, vetting, and payment graph data.
Monitor Claude Code costs, ROI, budgets, anomalies, and usage forecasts.