securevector-ai-monitor 4.1.0

Real-time AI threat monitoring. Protect your apps from prompt injection, leaks, and attacks in just a few lines of code.

SecureVector

Stop your AI agent burning $400 overnight.

Hard budget caps + auto-stop, per agent. Plus injection blocks, tool-call audits, and skill scanning — all in one local install. Free, open-source, runs on your machine. No signup. pip install and you're protected in 60 seconds.

Website · Getting Started · Verify your install · Discord · Dashboard Screenshots

▶ Watch the Demo

Threat detection, tool permissions, and cost tracking — running locally in real time.

What's new in v4.1.0:

• Agent Activity Timeline — local-first per-agent feed merging threat scans, tool-call audits, and LLM cost. Filter by agent + range, click to expand, CSV export. Traces never leave the box.
• Indirect Prompt Injection (IDPI) module — new direction="incoming" scan mode catches injections hidden in fetched RAG / HTML / email content. Ships with a 12-rule starter pack (zero-width unicode, hidden HTML comments, role overrides, tool-call hijack, exfil URLs, base64 decode-and-execute, credential exfil, javascript: markdown URIs, …).
• Build-provenance attestations on every wheel — SLSA Build Level 2+ via Sigstore Fulcio, anchored in Rekor. Verify with gh attestation verify. See SECURITY.md.
• Per-agent filter on Threat Monitor — Threats page joins Cost Tracking in being sliceable by agent.

Already in v4.0.0+: SIEM Forwarder, tool-call audit hash chain, per-device ID, OpenClaw Plugin, Skill Scanner, Tool Permissions, Cost Tracking & Budget Limits.

How It Works

SecureVector protects your AI agents at three layers:

• Pre-install — the Skill Scanner analyzes agent skill packages for shell access, network calls, and hidden risks before you install them
• Runtime — scans every prompt, response, and tool call for injection attacks, data leaks, and unauthorized access
• Observe — the SIEM Forwarder ships every threat + tool-call audit to your SOC in OCSF 1.3.0 format (Splunk HEC, Datadog, Microsoft Sentinel, Google Chronicle, IBM QRadar, OTLP, generic webhook, or a local NDJSON file) so AI events correlate with your existing security signals. Metadata-only by default; raw data is opt-in per destination.

For OpenClaw, the native plugin runs inside the agent with zero latency. For other frameworks, the multi-provider proxy intercepts traffic. 100% local — events only leave the machine when you configure a SIEM destination you control.

The Problem	The Fix
AI agents are powerful — and completely unprotected. Every prompt your AI agent sends, every secret it handles, every piece of user data — goes straight to the LLM provider with nothing in between. No spend limit. No injection protection. No audit trail. You're flying blind.	SecureVector runs on your machine. For OpenClaw/ClawdBot, the native plugin handles everything — zero latency, no proxy overhead. For LangChain, CrewAI, and other frameworks, the multi-provider proxy routes traffic across OpenAI, Anthropic, Ollama, and more. It blocks threats, enforces tool permissions, and hard-stops agents that blow their budget. 100% local. No accounts.

Quick Start

Step 1 — Install or download

pip install securevector-ai-monitor[app]
securevector-app --web

Or download the app: Windows · Linux · DEB · RPM · macOS (signed binary coming soon)

Step 2 — Open the app

Open http://localhost:8741 in your browser, or double-click the installed binary.

Step 3 — Connect your agent

OpenClaw / ClawdBot (plugin, zero latency)	LangChain, CrewAI, Ollama, n8n (proxy)
Observability & Monitoring — Go to Integrations → OpenClaw, click Install Plugin, restart OpenClaw. Done. No proxy, no env vars.	Observability & Monitoring — Go to Integrations, pick your framework, click Start Proxy, and set the env var shown on the page.

Block Mode (only if you want to enforce blocking) — Toggle Block Mode on the dashboard. The proxy starts automatically and blocks threats before they reach the LLM. Adds ~10–50ms latency per request. Applies to both plugin and proxy integrations.

If the app fails to launch because ports 8741/8742 are already in use, use --port <port> of your choice — the proxy starts automatically on port+1. See Configuration for proxy or web/api port settings.

Open-source. 100% local by default. No API keys required.

Screenshots

All screenshots are from a local app instance.

Tool Call History — 305 calls, 158 blocked: bash rm -rf, gmail_send to attacker, use_aws_cli stopped	Tool Permissions — allow or block tools by name or category	Tool Call Detail — decision, tool, args, and timestamp for every call
Dashboard — threat counts, cost metrics, and tool permission status	LLM Cost Tracker — per-agent spend, budgets, and token breakdown	Custom Rules — create and manage detection rules by category and severity
Skill Scanner — static security analysis for AI agent skills with scan history and risk levels	Skill Policy — network permissions, trusted publishers, and policy thresholds

What You Get

Threat Protection	Cost Control
Scans every prompt and response for prompt injection, jailbreaks, PII leaks, and tool abuse. 50+ detection rules covering the OWASP LLM Top 10. Monitors and alerts by default with zero latency (plugin mode) — enable block mode when you're ready to hard-stop threats via proxy.	Tracks every token and dollar per agent in real time. Set daily budget limits — requests auto-stop when the cap is hit. Never wake up to a surprise bill.
Skill Scanner	Full Visibility
Scan agent skills and tool packages before installing. Static analysis across 10 categories detects shell access, network calls, env var reads, and more. Optional AI review filters false positives automatically.	Live dashboard showing every LLM request, tool call, token count, and threat event. See exactly what your agents are doing.
100% Local
Runs entirely on your machine. No accounts. No cloud. No data leaves your infrastructure. Open source under Apache 2.0.

Features

Section	Feature	Description
Monitor	Threat Monitor	Live feed of every detected threat — prompt injection, jailbreaks, data leaks, tool abuse
	Tool Activity	Full audit log of every tool call your agents make, with args, decision, and timestamp
	Cost Tracking	Per-agent, per-model token spend and USD cost in real time, with request history
Scan	Skill Scanner	Static analysis of AI agent skills — detects shell exec, network access, env var reads, code injection, and 6 more categories
	AI Review	Optional LLM-powered false-positive filtering — works with OpenAI, Anthropic, Ollama, Azure, Bedrock
	Scan Policy	Risk scoring with per-category allow/block rules, trusted publishers, and severity thresholds
Configure	Tool Permissions	Allow or block specific tools by name or category — per agent, per rule. How allow / block / log_only are decided: see Tool Permissions guide
	Cost Settings	Set daily budget limits and choose whether to warn or hard-block at the cap
	Rules	Custom detection rules — auto-block or alert on threats matching your criteria

Performance: Rule-based analysis (default) adds ~10–50ms per request. Enabling optional AI analysis adds 1–3s per request depending on the model and provider — this is shown on the dashboard so you can measure it against your actual traffic.

Why SecureVector?

❌ Without SecureVector	✅ With SecureVector
Prompt injections pass straight through	Detected and alerted by default (zero latency); blocked when you enable block mode
API keys and PII leak in prompts	Automatically redacted
No control over what tools agents can use	Fine-grained allow/block rules per tool
No audit trail of tool calls	Full tool call history with decisions and reasons
No idea what agents are spending	Real-time cost tracking per agent
Runaway agents burn through your API budget overnight	Hard budget limits with auto-stop
Zero visibility into agent traffic	Live dashboard showing everything

Works With Everything

Your AI Stack

LangChain · LlamaIndex · CrewAI · AutoGen · LangGraph · n8n · Dify · OpenClaw/ClawdBot (LLM gateway agent framework) — or any framework that makes HTTP calls to an LLM provider.

LLM Providers

OpenAI · Anthropic · Ollama · Groq · and any OpenAI-compatible API.

Run Anywhere

Environment	Details
Local	macOS, Linux, Windows
Cloud	AWS, GCP, Azure
Containers	Docker & Kubernetes
Virtual Machines	EC2, Droplets, VMs
Edge / Serverless	Lambda, Workers, Vercel

Agent Integrations

Agent/Framework	Integration
LangChain	LLM Proxy or SDK Callback
LangGraph	LLM Proxy or Security Node
CrewAI	LLM Proxy or SDK Callback
Any OpenAI-compatible	LLM Proxy — see Integrations in UI
OpenClaw / ClawdBot (LLM gateway agent)	Native plugin (zero latency) — proxy only for block mode
n8n	Community Node
Claude Desktop	MCP Server Guide
Any OpenAI-compatible app	LLM Proxy — set OPENAI_BASE_URL to proxy
Any HTTP Client	POST http://localhost:8741/analyze with {"text": "..."}

OpenClaw / ClawdBot

Native plugin with ZERO latency — runs inside the agent, no proxy needed. Install from the Integrations tab or curl -X POST http://localhost:8741/api/hooks/install. Enable block mode from the dashboard when you want to actively stop threats via proxy.

Full setup guide

What It Detects

Input Threats (User to LLM)	Output Threats (LLM to User)
Prompt injection	Credential leakage (API keys, tokens)
Jailbreak attempts	System prompt exposure
Data exfiltration requests	PII disclosure (SSN, credit cards)
Social engineering	Jailbreak success indicators
SQL injection patterns	Encoded malicious content
Tool result injection (MCP)	—
Multi-agent authority spoofing	—
Permission scope escalation	—

Full coverage: OWASP LLM Top 10

AI Agent Attack Protection (28 new rules · 72 total)

Built from real attack chains observed against production agent frameworks:

• Tool Result Injection — injected instructions hidden inside MCP tool responses
• Multi-Agent Authority Spoofing — impersonating trusted agents in multi-agent pipelines
• Permission Scope Escalation — agents requesting more permissions than granted
• MCP Tool Call Injection — malicious payloads delivered through MCP tool calls
• Evasion techniques (22 rules) — zero-width characters, encoding tricks, roleplay framing, leetspeak, semantic inversion, emotional manipulation, and more

Device Identity

Every scan and audit row is stamped with a stable device_id so a customer running SecureVector across several laptops or agents can answer "which agent blocked this, which laptop is tampered, which machine spent what?" — not just "one of my installs did this".

Why we need it. A solo developer runs one install. A SOC team runs five to fifty. When an audit chain breaks, or a spike of blocked gmail-send calls shows up, the useful first question is which machine. Without a per-device tag, the answer is "some install" — which is useless in a fleet. device_id pins every row to a specific machine so dashboards, alerts, and compliance reviews can slice by device.

How it's generated (src/securevector/app/utils/device_id.py):

1. Read the OS's existing stable machine identifier:
   • macOS → IOPlatformUUID via ioreg
   • Linux → /etc/machine-id (fallback /var/lib/dbus/machine-id)
   • Windows → HKLM\SOFTWARE\Microsoft\Cryptography\MachineGuid
2. SHA-256 hash it with a namespace prefix (securevector-device-v1:<raw>) and truncate to 24 hex chars → sv-a1b2c3d4e5f6...
3. Cache the result in ~/Library/Application Support/ThreatMonitor/.device_id (0o600) so the OS fetch happens once per install.
4. If the OS refuses (rare: locked-down container, unusual Linux image), fall back to a random UUID cached to the same file.

Stability across reinstalls. The OS identifier outlives the app install — so uninstalling and reinstalling SecureVector on the same machine gives you the same device_id. Wiping the app data dir AND having no readable OS ID is the only combination that generates a new one. A new physical machine always gets a new ID.

Security / privacy posture — what the customer should know:

Concern	Reality
Is the raw OS machine UUID transmitted?	No. It's read locally, SHA-256 hashed with a namespace, and only the hash is stored. The raw value never reaches a log file or outbound event.
Can device_id be reversed to the OS UUID?	SHA-256 is one-way. An attacker who already has the raw OS UUID can compute the device_id — but they already have the machine at that point, so there's no incremental leak.
Does it track users?	No. It tracks machines. Multiple users on one laptop share one device_id. It's not tied to email, username, or any identity field.
Is it sent to SecureVector Cloud?	Only if Cloud Connect is on AND you trigger an action that reaches the cloud (rule sync, cloud-routed /analyze). device_id goes in metadata alongside scan results. You can opt out by keeping Cloud Connect off — local-only operation never transmits it.
Is it in SIEM forwards?	Yes, when the v4.1.0 SIEM forwarder is enabled — travels inside each OCSF event's unmapped block so your Splunk/Datadog can group by device.
Can the customer reset it?	Yes — delete .device_id in the app data dir. Next write will regenerate from the OS identifier (so same ID reappears) OR a fresh random UUID if the OS ID is unavailable.
Does it collide across containers cloned from the same image?	Potentially yes (they share /etc/machine-id). Not relevant for desktop use; mention it if you're deploying in Kubernetes.

In one sentence: device_id is a machine-identifier-per-install, derived locally, hashed before storage, never transmitted except with explicit user opt-in (Cloud Connect or SIEM Forwarder).

SIEM Forwarder

Stream every threat detection and tool-call audit into your own SIEM — Splunk HEC, Datadog, Microsoft Sentinel, Google Chronicle, IBM QRadar, an OpenTelemetry collector, a local NDJSON file, or any HTTPS endpoint that accepts JSON. Your data, your pipes.

Why this is safe to ship with zero monetization:

Feature	What leaves your machine
Scan verdict	scan_id, verdict, threat_score, risk_level, detected_types[], counts, durations
Tool-call audit	seq, action, risk, prev_hash, row_hash (the chain witness — lets your SIEM verify integrity)
Never transmitted	Prompt text, LLM output, matched patterns, reviewer reasoning, model reasoning

The allow-list is enforced at enqueue time by _assert_metadata_only(). Even if the forwarder code were tampered with, it can't add the forbidden fields back.

Supported destinations (one code path, OCSF 1.3.0 payload):

Kind	Target	Auth header
splunk_hec	https://<host>/services/collector/event	Authorization: Splunk <HEC-token>
datadog	https://http-intake.logs.<site>/api/v2/logs	DD-API-KEY: <key>
otlp_http	https://<collector>/v1/logs	optional Authorization: Bearer <token>
webhook	anything that accepts JSON POST	optional Authorization: Bearer <token>

Configure in Connect → SIEM Forwarder. Add SIEM destination → pick type → paste URL + token → Test → Save. Tokens are stored 0o600 in the app data dir, never in SQLite.

📊 Starter dashboards included:

Platform	Template
Microsoft Sentinel	docs/siem/sentinel/securevector-workbook.json
Splunk	docs/siem/splunk/securevector-dashboard.xml
Datadog	docs/siem/datadog/securevector-dashboard.json
Grafana (Loki)	docs/siem/grafana/securevector-dashboard.json

Each carries severity counters, events-over-time by severity, actor and MITRE-ish breakdowns, and a recent-high-severity log feed. MIT-licensed, AS-IS. Full install steps + field reference in docs/siem/README.md; trademark + upstream licenses in docs/siem/NOTICE.

Starter templates — import-test in your own stack and adjust queries / facets / sourcetypes before relying on them for production detections.

Reliability:

• Per-destination outbox with at-least-once delivery.
• A failing Datadog destination never blocks a healthy Splunk one.
• Per-destination circuit breaker backs off broken endpoints (1 min → 1 hour cap).
• Rows that fail 10 times are dropped (the health view shows the consecutive-failure count).

SIEM-side integrity verification. Every forwarded tool-call audit row carries its prev_hash and row_hash. Run a nightly search in your SIEM that rebuilds the chain — if a historic row has been tampered with on the local host, the forwarded evidence still tells the true story. That's the actual tamper evidence; the local chain alone is only the low bar.

Skill Scanner

Scan AI agent skills and tool packages before you install them. SecureVector performs static analysis across 10 detection categories, assigns a risk score, and optionally runs an AI review to filter false positives.

┌──────────────────────────────────────────────────────────────────────┐
│                        Skill Scanner Flow                           │
│                                                                     │
│   ┌─────────────┐     ┌──────────────────┐     ┌────────────────┐  │
│   │  Skill Dir   │────>│  Static Analysis  │────>│  Risk Scoring  │  │
│   │  or URL      │     │  (10 categories)  │     │  LOW/MED/HIGH  │  │
│   └─────────────┘     └──────────────────┘     └───────┬────────┘  │
│                                                         │           │
│                              ┌───────────────────────── │           │
│                              v                          v           │
│                     ┌─────────────────┐     ┌────────────────────┐  │
│                     │  AI Review      │     │  Policy Engine     │  │
│                     │  (optional LLM) │     │  allow/block rules │  │
│                     │  FP filtering   │     │  trusted publishers│  │
│                     └────────┬────────┘     └─────────┬──────────┘  │
│                              │                        │             │
│                              v                        v             │
│                     ┌──────────────────────────────────────┐        │
│                     │  Verdict: PASS / WARN / BLOCK        │        │
│                     │  + detailed findings per category     │        │
│                     └──────────────────────────────────────┘        │
└──────────────────────────────────────────────────────────────────────┘

Detection Categories

Category	What It Finds
shell_exec	Subprocess calls, system commands
network_domain	HTTP requests, socket connections, DNS lookups
env_var_read	Access to environment variables (API keys, secrets)
code_exec	eval, dynamic code generation
dynamic_import	Runtime module loading
file_write	Writing to disk outside expected paths
base64_literal	Obfuscated payloads in base64 strings
compiled_code	.pyc, .so, .dll binaries embedded in the skill
symlink_escape	Symlinks pointing outside the skill directory
missing_manifest	No permissions.yml declaring required capabilities

AI-Powered Review

Enable AI analysis (OpenAI, Anthropic, Ollama, Azure, or Bedrock) to automatically review findings and filter false positives. The AI examines each finding in context and adjusts the risk level — reducing noise without hiding real threats.

Open Source

SecureVector is fully open source. No cloud required. No accounts. No tracking. Run it, fork it, contribute to it.

Built for solo developers and small teams who ship AI agents without a security team or a FinOps budget. If you are building with LangChain, CrewAI, OpenClaw, or any agent framework — and you do not have someone watching your agent traffic and API spend — SecureVector is for you.

Install

Option 1: pip

Requires: Python 3.9+ (MCP requires 3.10+)

pip install securevector-ai-monitor[app]
securevector-app --web

Option 2: Binary installers

No Python required. Download and run.

Platform	Download
Windows	SecureVector-v4.1.0-Windows-Setup.exe
macOS	SecureVector-4.1.0-macOS.dmg (signed binary coming soon)
Linux (AppImage)	SecureVector-4.1.0-x86_64.AppImage
Linux (DEB)	securevector_4.1.0_amd64.deb
Linux (RPM)	securevector-4.1.0-1.x86_64.rpm

All Releases · SHA256 Checksums

Security: Only download installers from this official GitHub repository. Always verify SHA256 checksums before installation. SecureVector is not responsible for binaries obtained from third-party sources.

macOS binary note: If you downloaded a previous .dmg release and macOS blocks it, we recommend installing via pip instead: pip install securevector-ai-monitor[app]. A signed macOS binary is coming soon. If you must use the .dmg, only download from this official GitHub repository, verify the SHA256 checksum, then run xattr -cr /Applications/SecureVector.app in Terminal.

Other install options

Install	Use Case	Size
pip install securevector-ai-monitor	SDK only — lightweight, for programmatic integration	~18MB
pip install securevector-ai-monitor[app]	Full app — web UI, LLM proxy, cost tracking, tool permissions	453 KB wheel · ~16 MB total on disk (incl. dependencies)
pip install securevector-ai-monitor[mcp]	MCP server — Claude Desktop, Cursor	~38MB

Configuration

SecureVector writes svconfig.yml to your app data directory on first run with sensible defaults.

# SecureVector Configuration
# Changes take effect on next restart.
# The config path is printed to the console when you start the app.
#
# Linux:   ~/.local/share/securevector/threat-monitor/svconfig.yml
# macOS:   ~/Library/Application Support/SecureVector/ThreatMonitor/svconfig.yml
# Windows: %LOCALAPPDATA%/SecureVector/ThreatMonitor/svconfig.yml

server:
  # Web UI / API server listen host and port.
  # Change these if port 8741 is already in use on your machine.
  # If running on a remote server, set host to the server's hostname or IP address.
  host: 127.0.0.1
  port: 8741

security:
  # Block detected threats (true) or log/warn only (false)
  # Defaults to false — enable when you're confident in your rule tuning
  block_mode: false
  # Scan LLM responses for data leakage and PII
  output_scan: true

budget:
  # Daily spend limit in USD (set to null to disable)
  daily_limit: 5.00
  # Warn in logs/headers when spend approaches the limit
  warn: true
  # Block requests when the daily budget is exceeded
  block: true

tools:
  # Enforce tool permission rules (allow/block based on your rules)
  enforcement: true           # default: true

proxy:
  # OpenClaw/ClawdBot: proxy only starts when block_mode is enabled (above).
  #   Plugin-only mode handles monitoring with zero latency — no proxy needed.
  # LangChain/CrewAI/Ollama/other: proxy auto-starts as the only integration path.
  integration: openclaw       # or: langchain, langgraph, crewai, ollama
  mode: multi-provider        # or: single (add provider: below)
  provider: null              # required only when mode is "single"
  host: 127.0.0.1             # proxy listen host — set to the server's hostname or IP if running remotely
  port: 8742                  # proxy listen port (default: server.port + 1)

The UI keeps this file in sync — changes in the dashboard are written back to svconfig.yml automatically.

Pointing Your Agent at the Proxy

For LangChain, CrewAI, Ollama, and other non-OpenClaw frameworks, point your application to SecureVector's proxy instead of the provider's API. OpenClaw/ClawdBot users only need this when block mode is enabled.

🪟 Windows

🐧 Linux / macOS

Command Prompt (current session) set OPENAI_BASE_URL=http://localhost:8742/openai/v1 set ANTHROPIC_BASE_URL=http://localhost:8742/anthropic PowerShell (current session) $env:OPENAI_BASE_URL="http://localhost:8742/openai/v1" $env:ANTHROPIC_BASE_URL="http://localhost:8742/anthropic" PowerShell (persistent, per user) [Environment]::SetEnvironmentVariable( "OPENAI_BASE_URL", "http://localhost:8742/openai/v1", "User" )

Terminal (current session) export OPENAI_BASE_URL=http://localhost:8742/openai/v1 export ANTHROPIC_BASE_URL=http://localhost:8742/anthropic Persistent (add to ~/.bashrc or ~/.zshrc) echo 'export OPENAI_BASE_URL=http://localhost:8742/openai/v1' >> ~/.bashrc echo 'export ANTHROPIC_BASE_URL=http://localhost:8742/anthropic' >> ~/.bashrc source ~/.bashrc

Every request is scanned for prompt injection. Every response is scanned for data leaks. Every dollar is tracked — whether via native plugin (OpenClaw) or proxy (all other frameworks).

Supported providers (13): openai anthropic gemini ollama groq deepseek mistral xai together cohere cerebras moonshot minimax

Update

Method	Command
PyPI	pip install --upgrade securevector-ai-monitor[app]
Source	git pull && pip install -e ".[app]"
Windows	Download latest .exe installer and run it (overwrites previous version)
macOS	Download latest .dmg, drag to Applications (signed binary coming soon)
Linux AppImage	Download latest .AppImage and replace the old file
Linux DEB	sudo dpkg -i securevector_<version>_amd64.deb
Linux RPM	sudo rpm -U securevector-<version>.x86_64.rpm

After updating, restart SecureVector.

Documentation

• Installation Guide — Binary installers, pip, service setup
• Use Cases & Examples — LangChain, LangGraph, CrewAI, n8n, FastAPI
• MCP Server Guide — Claude Desktop, Cursor integration
• API Reference — REST API endpoints
• Security Policy — Vulnerability disclosure

Contributing

git clone https://github.com/Secure-Vector/securevector-ai-threat-monitor.git
cd securevector-ai-threat-monitor
pip install -e ".[dev]"
pytest tests/ -v

Contributing Guidelines · Code of Conduct

License

Apache License 2.0 — see LICENSE.

The starter SIEM dashboard templates under docs/siem/ (Splunk XML, Sentinel workbook, Datadog + Grafana JSON) are MIT-licensed — see docs/siem/LICENSE and docs/siem/NOTICE for trademark disclaimers.

SecureVector is a trademark of SecureVector. See NOTICE.

---

Get Started · Documentation · Discord · GitHub Issues · [e-mail]