Agent safety guardrail audit — Atom
MCP serverDeterministic API/MCP check that scores an agent's declared safety guardrails
Not checked yet- Remote
- Security
Deterministic API/MCP check that scores an agent's declared safety guardrails
Not checked yet45 pay-per-call agent tools for security, crypto, FX and web checks in one endpoint
Not checked yetReputation, CVE and prompt-injection checks for IPs, URLs, hashes and email
Not checked yetFind hidden Unicode, homoglyphs and lookalike names in text before an agent acts on it
Not checked yetPrompt-injection scanner for text, web pages and MCP tool definitions, paid per call
Not checked yetAudits MCP and A2A endpoints for schema changes, prompt injection and tool quality
Not checked yetRemote MCP package of 12 pre- and post-execution security checks for autonomous agents
Not checked yetAsk an action firewall for an allow, review or block verdict before an agent acts
Not checked yetScan MCP tool definitions for poisoning, check a server's change history, count token cost
Not checked yetTyped, policy-checked server operations for AI agents, with approvals and an audit log
Runs locallyScan untrusted text for hidden prompt injection: Unicode tags, zero-width, homoglyphs
Not checked yetRead your Golem agent fleet sessions, live activity and signed record from any client
Not checked yetThreat-intelligence verdicts on a URL, domain or IP before a coding agent connects
Not checked yetMCP server giving AI agents pre-action security checks for code, packages, links
Runs locallyOffline security inspection for MCP configs, metadata and dependencies, run from the CLI
Runs locallyMCP server that gives AI agents real inboxes with send, receive and injection scanning
Not checked yetPins MCP tool definitions in a lockfile and flags rug-pulls, poisoning and leaked secrets
Runs locallyPay-per-call prompt-injection screening and sanitising API for AI agents, via x402
Not checked yetRead-only MCP server that lints and budgets CLAUDE.md, AGENTS.md and other steering files
Runs locallyScreen documents for edits, failed redactions and prompt injection before agents read them
Runs locallyScore text and web pages for signs of AI writing, and check images for hidden instructions
Not checked yetScore a website for AI-agent safety, llms.txt and machine-readable signals
Runs locallyRead-only scam checks for messages, npm packages and crypto addresses before you act
Runs locallyFind and strip zero-width, bidi and tag characters hidden in text, offline
Runs locally
34 agent tools: free prompt-injection scans and checks plus pay-per-call data via x402
Not checked yetPayment firewall that screens x402 agent payments and returns allow, flag or block
Runs locally