supership-scan
Predeploy security scanner for AI code. 80+ patterns. Runs locally. x402 attestation.
supership-scan
Predeploy security scanner for the agent economy. Built by Crest Deployment Systems.
Scans your code for 80+ vulnerability patterns across secrets, auth, injection, config, Supabase, and logging. Runs locally. Your code never leaves the machine.
Install
npm install -g supership-scan
Requires Node.js 18+.
Usage
CLI
supership-scan .
Scans the current directory and prints findings.
supership-scan ./my-project --attest
Scans and requests a witnessed attestation ($0.01 USDC on Base). Only the report envelope (hashes and findings) is transmitted. Never source code.
MCP Server
supership-mcp
Starts an MCP server for AI editors (Claude Code, Cursor, Windsurf). Exposes the scanner as tools that agents can call directly.
Example Output
supership v1.0.0
Scanning 42 files...
Score: 87/100
Grade: B
Findings:
HIGH AUTH-003 Missing auth middleware on /api/admin src/routes/admin.js:14
MEDIUM CFG-002 CORS wildcard in production src/server.js:8
LOW LOG-001 Error stack in response body src/middleware/error.js:22
Scan complete. Code never left this machine.
Rule Categories
| Category | Patterns | Examples |
|---|---|---|
| Secrets | 30+ | API keys, credentials, .env exposure, private keys |
| Auth | 12+ | Missing middleware, inverted logic, RLS gaps |
| Injection | 15+ | SQL interpolation, XSS, eval(), command injection |
| Config | 10+ | CORS wildcards, source maps, insecure cookies |
| Supabase | 8+ | RLS disabled, permissive policies, service_role misuse |
| Logging | 6+ | Sensitive data in logs, error stack exposure |
Scoring
Score starts at 100. Penalties: critical (-25), high (-10), medium (-5), low (-1).
Severity gates override the score:
- Any critical finding = grade F
- Any high finding = grade C max
| Grade | Score |
|---|---|
| A | 90+ |
| B | 75-89 |
| C | 60-74 |
| D | 40-59 |
| F | <40 or any critical |
Attestations
The scan is free. The attestation costs $0.01.
When you run --attest, supership sends a report envelope to the attestation server. The envelope contains hashes and findings only. The server signs it, anchors the hash to the chain, and returns a witnessed attestation.
The attestation proves a specific scan occurred at a specific time with specific results. It does not certify that code is secure.
What's transmitted: input hash, rule pack hash, engine version, findings, score, grade.
What's never transmitted: source code, file contents, environment variables.
Benchmark
npm test
Runs 20 deliberately vulnerable fixtures against the scanner. Expected: 90% true positive rate, 0 harmful false positives.
Privacy
- Scanning is entirely local. No network calls during a scan.
- Attestation transmits hashes and findings only. Never source code.
- No telemetry. No analytics. No tracking.
API
supership also runs as an x402-native API. Pay per scan with USDC on Base. No API keys, no subscriptions.
| Endpoint | Method | Price | Description |
|---|---|---|---|
/check | GET | Free | Trust check for any x402 service URL |
/scan/free | POST | Free | Score + grade, all 6 categories |
/scan/quick | POST | $1 | Secrets + config findings |
/scan/full | POST | $5 | All categories + fixes |
/scan/deep | POST | $15 | Full + LLM contextual review |
/attest | POST | $0.01 | Sign and witness a scan result |
API base: https://supership.crestsystems.ai
Discovery endpoints: agent.json | llms.txt | OpenAPI
Crest x402 Services
supership is part of the Crest Deployment Systems x402 service fleet. All services accept USDC payments on Base mainnet via the x402 protocol.
| Service | What it does | URL |
|---|---|---|
| supership | Predeploy security scanner + attestation | supership.crestsystems.ai |
| data | Crypto market data, token lookups, gas prices | data.crestsystems.ai |
| audit | Smart contract audit, code security, wallet risk | audit.crestsystems.ai |
Links
- supership API
- Documentation
- npm: supership-scan
- npm: @crestdeploymentsystems/supership-mcp
- Crest Deployment Systems -- deploying scalable intelligence
License
Apache 2.0. See LICENSE for details.
Rule engines (src/rules/) are Apache 2.0 with a relicense notice. See LICENSE for the full NOTICE.
Похожие серверы
Alpha Vantage MCP Server
спонсорAccess financial market data: realtime & historical stock, ETF, options, forex, crypto, commodities, fundamentals, technical indicators, & more
AppControl Windows System History
An MCP server that gives AI agents like Claude, Cursor, and Gemini access to historical Windows CPU, GPU, temperature, and privacy data gathered by the AppControl Windows app.
Sequential Thinking Multi-Agent System (MAS)
An MCP agent that utilizes a Multi-Agent System (MAS) for sequential thinking and problem-solving.
Together AI Image Server
A TypeScript-based server for generating images using the Together AI API.
RapidTools Validate
Deterministic JSON Schema validation with signed cryptographic attestations. Validate payloads, verify signatures, and retrieve attestation records.
MCP Java Bridge
A bridge for the MCP Java SDK that enables TCP transport support while maintaining stdio compatibility for clients.
Valyu
Access Valyu's knowledge retrieval and feedback APIs.
Refactory
Hybrid decomp tool — AI decides WHERE to split your monolith, deterministic engine COPIES the code. Minimize tokens, maximize syntax validity.
System Diagnostics
An MCP server for system diagnostics and monitoring on Ubuntu using common command-line tools.
Claude Project Memory MCP
Manages project memory and implementation logs for Claude Code using local project files.
Tolgee
Manage your app’s translations directly from your AI coding assistant. Search keys, create translations, trigger machine translation. All without leaving your editor.