A safety layer for autonomous DeFi agents - MCP server that provides contract scanning, transaction simulation, token safety checks, and risk assessment tools to protect AI agents from malicious contracts and scams before executing on-chain transactions.
Aegis provides 6 well-named, security-focused tools with clear descriptions and complete input schemas. All tools start with action verbs (scan_, simulate_, check_, assess_, trace_, search_) and descriptions are detailed (150-250 chars), explaining WHAT, WHEN, and WHY. Input parameters use Zod validation with regex patterns for addresses and proper type constraints. However, output schemas are not documented, responses are JSON stringified without explicit field definitions, forcing LLMs to infer structure. Error handling is minimal (only missing-param checks visible). No tool annotations (readOnlyHint, destructiveHint) despite all tools being read-only operations. Parameter descriptions are strong but could benefit from explicit constraint callouts (e.g., 'maxLength: 500000' should appear in description text, not just schema).
Comprehensive risk assessment combining contract scanning, transaction simulation, and token checks. This is the recommended all-in-one safety check before any DeFi interaction. Returns a go/no-go recommendation.
Check if a token is safe to trade. Detects honeypot mechanics (can't sell), concentrated holdings, fake ownership renouncement, and other scam indicators. Use this before swapping into any unfamiliar token.
Analyze a smart contract's source code or bytecode for known exploit patterns, honeypot mechanics, rug-pull signals, and security vulnerabilities. Returns a risk score (0-100) and detailed findings. Use this BEFORE interacting with any unfamiliar contract.
Search 50K+ real-world audit findings from top security firms. Find known vulnerabilities and exploits that match your contract.
Simulate a transaction on a forked chain WITHOUT actually executing it. Detects reverts, abnormal gas usage, and other red flags. Use this to preview what will happen before sending a real transaction.
Output schemas not documented. Tools return JSON.stringify(result) without explicit field definitions. LLMs cannot plan downstream calls or extract structured data reliably.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). All 6 tools are read-only and should declare this via annotations for agent safety and optimization.
Minimal error handling. Only missing-parameter validation visible. No recovery guidance (e.g., 'Try search_solodit() with a different query' if no results). No error categorization (retryable vs fatal).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 65 | 2026-07-28+ | v2 |
Trace all internal calls and scan every contract touched by a transaction. Detects delegatecall attacks, reentrancy patterns, and other multi-contract exploits.
Parameter constraints (maxLength, pattern) not reflected in descriptions. LLMs cannot read JSON Schema, they rely on description text. E.g., 'source' param should say 'max 500KB' in description, not just schema.
No pagination or result limits documented for search_solodit. If 50K+ audit findings exist, returning all matches could exhaust context. Should cap results and offer pagination.