More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    BurnGuard MCP — Token Budget Firewall & Cost Telemetry

    1

    Pre-Execution Token Budget Firewall & Cost Telemetry Sentinel. Blocks runaway debugging loops, giant file ingestion, and surprise API invoices.

    $6

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    BurnGuard MCP — Token Budget Firewall & Cost Telemetry

    BurnGuard MCP — Token Budget Firewall & Cost Telemetry

    Example session with this skill installed

    Enforce a $2.00 hard session budget cap with BurnGuard. Block reading any lockfiles or dist directories, and track real-time token spend across Claude 3.5 Sonnet.

    • Read your context and instructions
    • Compiled the burnguard mcp

    I have activated BurnGuard MCP with a hard $2.00 session budget. The pre-execution firewall intercepted a 105k-token context bloat attempt and established spend telemetry.

    BurnGuard Cost Telemetry & Firewall Report

    <budget_telemetry currency="USD" model="claude-3-5-sonnet">
      <hard_cap limit="2.00" current_spend="0.28" remaining="1.72" />
      <bloat_interceptions count="2" tokens_saved="118500" money_saved="$0.355">
    
    • Blocked direct read on `package-lock.json` (~105,000 tokens)
    • Blocked direct read on `dist/bundle.min.js` (~13,500 tokens)
    
      </bloat_interceptions>
    </budget_telemetry>
    

    Session Spend Breakdown & Interceptions

    ActionFile / CommandToken CountCost (USD)Status
    read_filepackage-lock.json105,000$0.315BLOCKED (Exit Code 2)
    read_filedist/bundle.min.js13,500$0.040BLOCKED (Exit Code 2)
    read_file_surgicalsrc/billing.ts:1-40850$0.002PERMITTED
    run_commandnpm test billing2,100$0.006PERMITTED

    Next Steps

    1. Current spend: $0.28 / $2.00. Headroom remaining: $1.72.
    2. If cumulative spend reaches $2.00, BurnGuard will halt the agent with Exit Code 2.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    BurnGuard prevents autonomous coding agents from racking up catastrophic API bills ($15–$50+) during runaway debugging loops, recursive file reads, and bloated context dumps.

    Autonomous agents frequently incur massive token burns due to unconstrained execution patterns:

    1. Accidental Giant File Ingestion: An agent attempts to read a 15MB minified bundle or SQL dump, consuming 100k+ tokens in a single tool call.

    2. Infinite Debugging Bleed: An agent loops through failing tests 20 times, burning $2.50 per turn without making progress.

    3. No Cumulative Guardrail: Developers wake up to depleted API credits or surprise $500 monthly bills.

    4. Pre-Execution Token Estimation:

      • Computes expected input and output tokens and estimated dollar cost across major frontier models before tool execution.
    5. Hard Spending Cap Enforcement:

      • Halts agent tool loops (exit code 2) once a pre-configured session limit (e.g. $2.00 or $5.00) is reached.
    6. Bloat Path Interceptor:

      • Immediately blocks reading or listing huge build artifacts, lockfiles, or vendor directories (node_modules, vendor, dist, .git).
    7. Live Spend Telemetry:

      • Provides a real-time dashboard of session expenditure, tokens consumed, and remaining headroom.
    • estimate_token_cost: Calculates token counts and financial cost for text payloads and commands.
    • enforce_budget_cap: Checks current session burn against hard limits and blocks execution if exhausted.
    • inspect_payload_bloat: Scans file paths and command arguments for token bloat vectors.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Workflow Automation.

    Token estimation uses standard ~4 character-per-token heuristics calibrated across Claude 3.5 Sonnet, GPT-4o, and Gemini 1.5 Pro.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 19 days ago

    Software
    terminal

    Permissions

    No special permissions declared or detected

    Listed19 days ago

    What's inside

    Frequently Asked Questions

    BurnGuard returns Exit Code 2, halting the agent tool execution loop until you explicitly approve a budget extension.

    New in Workflow Automation