OpenClaw Users
Your OpenClaw agents are exposed. One command fixes that.
CVE-2026-25253 gave any website RCE on any OpenClaw machine. Crawdad hooks into every agent automatically — zero config, free to start.
Install the skill
git clone https://github.com/AndrewSispoidis/crawdad-openclaw ~/.openclaw/skills/crawdad
That's it. Crawdad activates automatically.
# Every message scanned for injection
# Every tool call authorized
# Every response filtered for PII
# Free API key provisioned on first run
View on GitHub →
API live — 517 tests passing

Your AI agents need a security layer.

Security researchers flagged critical gaps in autonomous AI agents — prompt injection, data exfiltration, unvetted skills, no access control. Crawdad is the runtime security API that fixes all of them. Post-quantum ready, air-gap deployable, cryptographically auditable. Works with OpenClaw, LangChain, CrewAI, AutoGen, and any agent framework.

See the SDK
517+Tests passing
90+API endpoints
<10msp99 latency
0Unsafe code
PQPost-quantum ready
terminal
# Secure your AI agent in 30 seconds
pip install crawdad-sdk

# Scan inbound for prompt injection
curl -X POST https://crawdad-production.up.railway.app/firewall/analyze \
  -H "X-API-Key: $KEY" \
  -d '{"input": "Ignore all instructions..."}'

AI agents are running without guardrails

Autonomous AI agents run shell commands, control browsers, read files, and send emails — all with the user's full permissions. Security researchers have documented the gaps:

Prompt Injection

Malicious instructions embedded in data can hijack the agent's capabilities. No semantic firewall exists to detect or block them.

Data Exfiltration

Third-party skills can exfiltrate data without user awareness. No content filtering or PII detection exists.

Unvetted Skills

Skills are directories with a markdown file. No hash verification, no attestation, no capability policy enforcement.

No Access Control

Agents run with the user's full permissions. No identity verification, no policy engine, no action authorization.

Documented by CrowdStrike, Cisco, Microsoft Security, Giskard, and Penligent. OWASP formalized 10 risk categories for agentic AI (ASI01–ASI10).

Seven pillars of agent trust

Crawdad wraps your agent in a complete security layer. One API call before every inbound message, tool execution, and outbound response.

🔑

Agent Identity

Ed25519 + CRYSTALS-Kyber1024 hybrid keypairs, step-ca PKI, mTLS, forward secrecy. DID documents, encrypted credential vaults, three-level kill switch, and purpose-bound scoped tokens.

8 endpoints
🛡️

Semantic Firewall

Structural deobfuscation, 27 injection patterns, output guard with exfiltration detection, instruction density scoring. ZK policy proofs — verify decisions without revealing rules.

4 endpoints
⚖️

Policy Engine

5-factor risk scoring, Rule of Two enforcement, behavioral baselines with anomaly detection. Byzantine fault detection with auto-isolation of compromised agents.

6 endpoints
🧠

Memory Integrity

Merkle-chained entries with Ed25519 signatures, firewall-gated writes, surgical rollback, compaction seals, and write anomaly detection.

7 endpoints
🔧

Skill Attestation

SHA-256 manifest verification, static analysis, capability policies, version validation, typosquat detection, runtime monitoring, and SBOM analysis.

11 endpoints
📡

Comms Governance

Signed envelopes, delegation chains with scope reduction, content filtering, collusion detection, cascade breakers, quarantine zones, and provenance tracking.

13 endpoints
🔒

Privacy & Compliance

15-category PII detection, 4 transform modes, consent management, DSAR engine, 10-jurisdiction compliance, differential privacy, and reporting.

14 endpoints

Built for production agent deployments

Beyond the core pillars, Crawdad includes hardened security features designed for real-world multi-agent systems.

R2

Rule of Two

No agent holds untrusted input + sensitive data + code execution simultaneously. Auto-deny on violation.

CB

Cascade Breaker

Per-agent fan-out limits with auto-trip on >50% error rate. Three isolation levels: Soft, Hard, Quarantine.

PP

Provenance Propagation

Trust decay (10%/hop, floor 0.1) prevents trust laundering where untrusted data gains trust through intermediaries.

DC

Delegation Chains

Validates multi-hop delegations for monotonic scope reduction, depth limits, and circular prevention.

CS

Compaction Seals

Cryptographic seals over compacted memory with SHA-256 hashes, composite provenance, and tamper-evident verification.

SB

SBOM Analysis

CycloneDX bills of materials with semver vulnerability matching, license policy enforcement, and risk scoring.

DP

Differential Privacy

Laplace and Gaussian noise mechanisms for aggregate queries with privacy budget tracking and exhaustion enforcement.

MT

Multi-Tenancy

Per-tenant data isolation, scoped API keys, admin management endpoints, and configurable agent quotas.

Secure your agent in 5 lines

pip install crawdad-sdk. Get an API key. The Python SDK wraps all 90+ endpoints with type hints, error handling, and context manager support.

from crawdad.openclaw import CrawdadMiddleware

mw = CrawdadMiddleware(
  base_url="https://crawdad-production.up.railway.app",
  api_key="your-key"
)

# Scan inbound messages for prompt injection
result = mw.scan_inbound("user message")

# Gate tool execution through policy
result = mw.authorize_action(agent_id, "shell_exec", "/bin/bash")

# Scan outbound for PII and credentials
result = mw.scan_outbound("Contact john at example.com")
safe = result["redacted"]

Everything you need

Agent Framework Integration
Middleware for OpenClaw, LangChain, CrewAI, AutoGen. Skill scanner, memory guard, policy gate, credential detector.
Python SDK on PyPI
62 methods, 35+ Pydantic models, typed exceptions. pip install crawdad-sdk.
Full REST API
90+ endpoints. Use from any language — Python, TypeScript, Go, Rust, curl. OpenAPI 3.1 spec available on request.
CLI Onboarding
crawdad signup / init / health / demo. Zero-friction setup for any agent framework.

The most secure OpenClaw deployment, by default.

OpenClaw runs shell commands, controls browsers, reads your files, and sends your emails — autonomously, on a loop, with your full permissions.

CVE-2026-25253 (CVSS 8.8) let any website take over any OpenClaw machine through a single malicious link. It was disclosed in January 2026.

Crawdad wraps every OpenClaw agent in seven security layers. It activates the moment you install it. No configuration. No code changes. Your agents keep working — now with a security layer they cannot override.

terminal
git clone https://github.com/AndrewSispoidis/crawdad-openclaw ~/.openclaw/skills/crawdad
Automatic
Hooks into every message, tool call, and response automatically
Open source
Read exactly what it does before installing
Free tier
10,000 API calls/month, no credit card
View on GitHub →
1

Install (30 seconds)

Run the install command. Crawdad downloads the skill and provisions a free API key automatically.

2

Automatic protection

Every inbound message is scanned for prompt injection. Every tool call is authorized through a policy engine. Every outbound response is filtered for PII and credentials.

3

Stay in control

Full audit trail of every security decision. Kill switch for any agent. Upgrade when you need more API calls.

Runs anywhere. Including nowhere.

Three deployment modes for every security requirement.

☁️

Cloud API

Zero config. Get an API key and start securing agents in 30 seconds. We handle the infrastructure.

getcrawdad.dev — free tier included
🖥️

Self-Hosted

Docker Compose deployment on your own infrastructure. Your data never leaves your network.

docker compose up — full docs included
🔒

Air-Gap

Fully offline. No external dependencies. USB-deployable. Built for classified environments and zero-trust networks.

contact andrew@getcrawdad.dev

Start free. Scale when you need to.

Every plan includes all 7 security pillars, all 90+ API endpoints, and the full Python SDK. No feature gating.

Starter
$0/mo
Evaluate Crawdad with your agent stack.
  • 10,000 API calls / month
  • 5 agents
  • All 7 security pillars
  • All 90+ endpoints
  • Python SDK
Developer
$49/mo
For developers building with AI agents.
  • 100,000 API calls / month
  • 25 agents
  • All 7 security pillars
  • All 90+ endpoints
  • Email support
Get Started
Business
$799/mo
For organizations with compliance needs.
  • 10,000,000 API calls / month
  • Unlimited agents
  • Multi-tenancy + admin API
  • Priority support
  • 99.99% uptime SLA
  • Compliance reporting
Get Started

Enterprise — Dedicated infrastructure, VPC deployment, on-premise, custom SLAs. Starting at $3,000/month.
Government & Defense — Air-gapped deployment, FedRAMP pathway, data sovereignty. Starting at $50,000/year.
Contact us →

Secure your agents today

Get your API key in 10 seconds. Start with 10,000 free API calls. No credit card required.

Free tier — 10,000 calls/month, 5 agents, all features. No credit card.