Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Prasasti: Agent Sentinel & Columnar Telemetry Engine

    2

    Specifically engineered for autonomous AI agents, it equips agent hosts with microsecond observability, recursive loop circuit breakers, and deterministic token budget guards.

    $19

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Prasasti: Agent Sentinel & Columnar Telemetry Engine

    Prasasti: Agent Sentinel & Columnar Telemetry Engine

    Example session with this skill installed

    Record telemetry for the 'cargo_test' step with 145ms latency and 2,500 tokens, then check if our active agent loop circuit breaker has been tripped within the last 5 minutes (max 5 repeats, 50,000 token ceiling).

    • Read your context and instructions
    • Compiled the prasasti

    Telemetry & Circuit Breaker Assessment

    The agent step invocation was committed to Prasasti's hardware-CRC32 Write-Ahead Log. The sliding-window circuit breaker evaluation completed in

    18 microseconds with zero runtime allocations.

    Telemetry DimensionValueThreshold / LimitCircuit Status
    Ingestion StatusRECORDEDHardware CRC32 ValidatedCommitted
    Monitored Actioncargo_testActive Agent ToolNormal
    Window Invocation Count3 repeats5 repeat maximumALLOW
    Cumulative Tokens7,500 tokens50,000 ceilingALLOW
    Mean Step Latency142.6 msMonitored SLANominal
    Circuit Breaker ActionALLOWSafe to ProceedClear
    {
      "status": "ALLOW",
      "reason": "OK",
      "symbol": "cargo_test",
      "repeat_count": 3,
      "max_repeats": 5,
      "total_tokens": 7500.0,
      "max_tokens": 50000.0,
      "mean_latency_ms": 142.6,
      "scanned_segments": 1,
      "skipped_segments": 0,
      "license_tier": "ENTERPRISE_PRO_V1"
    }
    

    Next Steps

    • Tool execution is authorized to proceed; no runaway recursive loop detected.
    • Telemetry will remain persisted across sessions in the columnar segment store.

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

    About this skill

    Overview

    Prasasti is a zero-allocation, high-throughput embedded columnar telemetry and time-series engine written in 100% pure Rust (#![forbid(unsafe_code)]). Specifically engineered for autonomous AI agents, it equips agent hosts with microsecond observability, recursive loop circuit breakers, and deterministic token budget guards.

    Core Capabilities

    1. Runaway Tool Loop Circuit Breaker: Evaluates microsecond sliding-window telemetry to immediately trip circuit breakers (LOOP_DETECTED) if an agent enters an infinite repetitive loop.
    2. Token Quota Enforcement: Tracks cumulative token consumption and trips QUOTA_EXCEEDED before budget exhaustion occurs.
    3. Hardware-CRC32 WAL Durability: Ingests agent telemetry into an append-only Write-Ahead Log capable of automatic corrupted-tail crash recovery.
    4. Vectorized Aggregations & OHLCV: Performs sub-millisecond range aggregations and extracts contiguous tensor patches for time-series forecasting foundation models (Google TimesFM 2.5 / Kronos).

    Enterprise Pro Commercial License Included

    This bundle includes the pre-configured PRASASTI_LICENSE_KEY inside mcp.json, cryptographically unlocking unlimited Enterprise Pro telemetry throughput and removing all Community Evaluation limits.

    Requires Windows x86_64 for the precompiled binary. Requires MCP-compatible agent host like Claude Desktop or Cursor.

    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 any payment.

    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 12 days ago

    Permissions

    Write Files
    Read Files

    File Scopes

    *.wal
    *.prs

    Prasasti records append-only Write-Ahead Log (.wal) and immutable columnar segments (.prs) locally on disk.

    Listed12 days ago

    What's inside

    Frequently Asked Questions

    Prasasti Sentinel acts as an external safety layer that monitors agent behavior in real-time. It uses a high-performance Rust engine to detect infinite tool-calling loops and token budget exhaustion, automatically signaling the agent to stop before costs spiral out of control.

    The skill is compatible with any AI agent host that supports the Model Context Protocol (MCP), including Cursor, Claude Desktop, and custom MCP integrations. The included binary is currently optimized for Windows x86_64 environments.

    Setup is straightforward: you register the provided mcp.json file with your preferred agent host. Once configured, your agent will be able to record telemetry and check circuit health during its standard tool-calling workflow.

    Unlike prompts, which agents can ignore during a logic loop, Prasasti operates outside the model's inference context. It uses a hardware-level Write-Ahead Log (WAL) and a dedicated columnar engine to enforce hard limits that the agent cannot 'hallucinate' its way around.

    You will receive the precompiled Windows x86_64 binary, the MCP server configuration file (mcp.json), and comprehensive documentation on integrating the telemetry and circuit-breaking tools into your agent's routine.

    New in LLM Ops