AI Bridge
Secure bridge to OpenAI and Google Gemini APIs with multi-layer security protection including prompt injection detection, content filtering, input sanitization, and configurable rate limiting to prevent API abuse while maintaining access to all major models.
0Tools
11Findings
3Stars
Mar 22, 2026Last Scanned
7 critical · 3 high · 1 low findings detected
Security Category Deep Dive
Prompt Injection
Prompt & context manipulation attacks
69
Maturity
14
Rules
5
Sub-Categories
1
Gaps
64%
Implemented
56
Tests
1
Stories
100%3 rules
Injection via tool descriptions and parameter fields
GAP-001Prompt Injection Coverage GapMissing detection coverage for emerging prompt injection attack variants not addressed by current rules
100%4 rules
Hidden instructions via external content and tool responses
100%2 rules
Context window saturation and prior-approval exploitation
100%3 rules
Payload hiding via invisible chars, base64, schema fields
100%2 rules
Injection via prompt templates and runtime tool output
Findings11
7critical
3high
1low
Critical7
criticalQ13MCP Bridge Package Supply Chain AttackMCP10-supply-chainAML.T0054
Pattern "["']@modelcontextprotocol/sdk["']\s*:\s*["'](?:\^|~|\*|latest)" matched in source_code: ""@modelcontextprotocol/sdk": "^" (at position 11094)
MCP bridge packages (mcp-remote, mcp-proxy, @modelcontextprotocol/sdk, fastmcp) are high-value supply chain targets — CVE-2025-6514 (CVSS 9.6) in mcp-remote affected 437,000+ installs. Always pin exact versions (no ^ or ~ ranges). Use lockfiles (package-lock.json, pnpm-lock.yaml, uv.lock). Never run `npx mcp-remote` without version pinning. Verify package integrity with `npm audit` or `pip-audit` before deployment. Reference: CVE-2025-6514, OWASP ASI04.
criticalL7Transitive MCP Server DelegationMCP06-excessive-permissionsAML.T0054
Pattern "(?:proxy|forward|delegate|relay|bridge).*(?:mcp|tool|request).*(?:server|client|upstream)" matched in source_code: "Bridge MCP server" (at position 10223)
MCP servers MUST NOT create client connections to other MCP servers without explicit user disclosure. If delegation is required, declare all downstream servers in the server's capabilities and tool descriptions. Never forward user credentials to sub-servers. Implement a trust boundary between the approved server and any delegated servers. Log all transitive delegations for audit.
criticalL9CI/CD Secret Exfiltration PatternsMCP07-insecure-configAML.T0057
Pattern "(?:writeFileSync|writeFile|open.*w).*(?:process\.env|os\.environ).*(?:TOKEN|SECRET|KEY|CREDENTIAL|PASSWORD)" matched in source_code: "openai = new OpenAI({ apiKey: process.env.OPENAI_API_KEY" (at position 2133)
Never print, log, or transmit CI environment variables containing secrets. Use GitHub Actions '::add-mask::' to prevent accidental secret exposure in logs. Audit all third-party Actions for secret access patterns. Use OIDC tokens instead of long-lived secrets where possible. Restrict secret access to specific workflow jobs and steps. Monitor CI logs for base64-encoded strings.
criticalQ1Dual-Protocol Schema Constraint LossMCP06-excessive-permissionsAML.T0054
Pattern "(mcp|model[_\s-]?context).{0,60}(openai|function[_\s-]?calling|tool[_\s-]?choice)" matched in source_code: "MCP server for bridging Claude Code with OpenAI" (at position 10769)
Dual-protocol MCP servers must validate input constraints server-side, not rely on client-side schema enforcement. When translating MCP schemas to OpenAI function-calling format, constraints like `pattern`, `minLength`, `maxLength`, and `format` are silently dropped. Implement server-side Zod/AJV validation on all tool inputs regardless of the calling protocol.
criticalC1Command InjectionMCP03-command-injectionAML.T0054
Pattern "`[^`]+`" matched in source_code: "`The model to use (default: ${DEFAULTS.OPENAI.MODEL})`" (at position 3604)
Replace exec()/execSync() with execFile() and pass arguments as an array, never as a string. Validate all inputs against an allowlist before use in any shell context. For subprocess.run, always pass a list and shell=False.
criticalJ7OpenAPI Specification Field InjectionMCP03-command-injectionAML.T0054
Pattern "(?:summary|operationId|description).*\$\{|`.*(?:summary|operationId)" matched in source_code: "description: `The model to use (default: ${" (at position 3591)
Sanitize all OpenAPI specification fields before using them in code generation or template interpolation. Treat summary, description, operationId, and extension fields as untrusted input. Use parameterized templates instead of string interpolation. See CVE-2026-22785/23947.
criticalK14Agent Credential Propagation via Shared StateMCP05-privilege-escalationAML.T0054
Pattern "(process\.env|os\.environ|setenv|putenv).*(?:token|credential|api[_\s-]?key|secret|password)" matched in source_code: "process.env.OPENAI_API_KEY" (at position 2016)
Never write credentials to shared agent state. Use credential vaults (HashiCorp Vault, AWS Secrets Manager) with per-agent scoped access. Implement OAuth token exchange (RFC 8693) for cross-agent authorization. Redact credentials from all agent outputs before writing to shared memory. Required by OWASP ASI03/ASI07 and MAESTRO L7.
High3
highD1Known CVEs in DependenciesMCP08-dependency-vuln
Dependency "@modelcontextprotocol/sdk@0.5.0" has known CVEs:
Update dependencies to versions that patch known CVEs. Run 'npm audit fix' or 'pip-audit' to identify and resolve vulnerable dependencies.
highK16Unbounded Recursion / Missing Depth LimitsMCP07-insecure-configAML.T0054
Pattern "(invoke|call|execute)[_\s-]?(?:tool|agent|self)(?!.*(?:depth|level|limit|max[_\s-]?(?:depth|recursi|iter|call)|count))" matched in source_code: "CallTool" (at position 190)
Add explicit depth/recursion limits to all recursive operations. Use iterative approaches where possible. Set maximum depth for directory walking (max_depth=10), tree traversal (max_level=20), and agent re-invocation (max_calls=5). Implement circuit breakers that halt after N iterations. Required by EU AI Act Art. 15 (robustness) and OWASP ASI08.
highK13Unsanitized Tool OutputMCP02-tool-poisoningAML.T0054
Pattern "(?:readFile|read_file|open).*(?:return|respond|result|content|text)(?!.*(?:sanitize|escape|encode|strip|filter|validate|truncate))" matched in source_code: "OPENAI RESPONSE (${model}):\n\n${completion.choices[0].message.content" (at position 7071)
Sanitize all external data before including in tool responses. Implement output encoding that neutralizes prompt injection patterns. Truncate excessively long content. Validate structure before passing database results. Apply the principle: treat all external data as untrusted, even in tool outputs. Required by CoSAI MCP-T4.
Low1
lowF4MCP Spec Non-ComplianceMCP07-insecure-config
Server fails MCP spec compliance checks: required:server_name; required:server_version; required:protocol_version; recommended:tool_descriptions; recommended:parameter_descriptions
Follow the MCP specification for server metadata. Include server name, version, and protocol version. Provide descriptions for all tools and parameters.