AIR SDK
Collective intelligence for browser automation agents. Site capabilities, selectors, and extraction.
0Tools
20Findings
1Stars
Mar 24, 2026Last Scanned
5 critical · 13 high · 1 medium · 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
50%2 rules1 found
Injection via prompt templates and runtime tool output
Findings20
5critical
13high
1medium
1low
Critical5
criticalL9CI/CD Secret Exfiltration PatternsMCP07-insecure-configAML.T0057
Pattern "(?:btoa|Buffer\.from|base64\.b64encode).*(?:process\.env|os\.environ|TOKEN|SECRET|KEY|PASSWORD)" matched in source_code: "Buffer.from(sessionKey" (at position 48107)
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.
criticalC1Command InjectionMCP03-command-injectionAML.T0054
Pattern "`[^`]+`" matched in source_code: "`apiKey must start with "${API_KEY_PREFIX}"`" (at position 24266)
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.
criticalQ13MCP Bridge Package Supply Chain AttackMCP10-supply-chainAML.T0054
Pattern "["']@modelcontextprotocol/sdk["']\s*:\s*["'](?:\^|~|\*|latest)" matched in source_code: ""@modelcontextprotocol/sdk": "^" (at position 50956)
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.
criticalK5Auto-Approve / Bypass Confirmation PatternMCP06-excessive-permissionsAML.T0054
Pattern "(auto|automatic)[_\s-]?(approv|confirm|accept|consent|yes)" matched in source_code: "autoApprov" (at position 5210)
Never auto-approve or bypass human confirmation for operations with side effects. Implement explicit confirmation gates that cannot be programmatically bypassed. If batch/CI mode is needed, require an explicit opt-in flag with audit logging. Required by EU AI Act Art. 14 (human oversight) and OWASP ASI09.
criticalK8Cross-Boundary Credential SharingMCP05-privilege-escalationAML.T0054
Pattern "(return|respond|output|result).*(?:token|credential|api[_\s-]?key|secret|password|bearer)" matched in source_code: "return apiKey" (at position 26295)
Never forward, share, or embed credentials across trust boundaries. Use OAuth token exchange (RFC 8693) to create scoped, delegated tokens instead of passing original credentials. Never include credentials in tool responses. Required by ISO 27001 A.5.17 and OWASP ASI03.
High13
highD1Known CVEs in DependenciesMCP08-dependency-vuln
Dependency "playwright@1.42.0" has known CVEs:
Update dependencies to versions that patch known CVEs. Run 'npm audit fix' or 'pip-audit' to identify and resolve vulnerable dependencies.
highD1Known CVEs in DependenciesMCP08-dependency-vuln
Dependency "tsup@8.0.0" has known CVEs:
Update dependencies to versions that patch known CVEs. Run 'npm audit fix' or 'pip-audit' to identify and resolve vulnerable dependencies.
highD1Known CVEs in DependenciesMCP08-dependency-vuln
Dependency "vitest@1.2.0" has known CVEs:
Update dependencies to versions that patch known CVEs. Run 'npm audit fix' or 'pip-audit' to identify and resolve vulnerable dependencies.
highK13Unsanitized Tool OutputMCP02-tool-poisoningAML.T0054
Pattern "(?:query|execute|select|find).*(?:return|respond|result|rows|data)(?!.*(?:sanitize|escape|encode|map|filter|select|pick))" matched in source_code: "ExecuteResult" (at position 921)
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.
highK16Unbounded Recursion / Missing Depth LimitsMCP07-insecure-configAML.T0054
Pattern "function\s+(\w+).*\{[^}]*\1\s*\((?!.*(?:depth|level|limit|max|count|recursi))" matched in source_code: "function extractApiKeyPrefix(apiKey: string): string {
return apiKey.slice(" (at position 26238)
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.
highK18Cross-Trust-Boundary Data Flow in Tool ResponseMCP04-data-exfiltrationAML.T0054
Pattern "(?:readFile|read_file|query|select|getSecret|getCredential|getPassword).*(?:webhook|http|fetch|axios|post|send|email|slack|discord)" matched in source_code: "Selectors = new Map<string, { data: SmartSelector; fetch" (at position 39132)
Implement data flow taint tracking: tag data from sensitive sources (databases, credentials, files) and prevent it from flowing to external sinks (HTTP, webhooks, email) without explicit sanitization/redaction. Apply data classification and enforce boundary controls per trust level. Required by ISO 27001 A.5.14 and CoSAI MCP-T5.
highK4Missing Human Confirmation for Destructive OpsMCP06-excessive-permissionsAML.T0054
Pattern "(auto[_\s-]?execute|auto[_\s-]?approve|skip[_\s-]?confirm|no[_\s-]?prompt)" matched in source_code: "autoApprove" (at position 5210)
All destructive operations (delete, drop, overwrite, send) MUST include a human confirmation step. Use the MCP destructiveHint annotation to signal that client-side confirmation is required. Implement an approval gate pattern: preview changes → request confirmation → execute. Required by ISO 42001 A.9.1, EU AI Act Art. 14, and NIST AI RMF GOVERN 1.7.
highQ14Concurrent MCP Server Race ConditionMCP07-insecure-configT1068
Pattern "(?:read|write|modify|delete).*(?:file|path|directory)(?!.*(?:lock|mutex|semaphore|flock|atomic))" matched in source_code: "writeFile" (at position 50517)
MCP servers sharing filesystem or database backends with other servers must implement proper concurrency controls. Use: (1) file locking (flock/lockfile) for filesystem operations, (2) database transactions for all read-modify-write sequences, (3) atomic file operations (O_EXCL, mkdtemp) instead of check-then-create, (4) lstat() to detect symlinks before following (CVE-2025-53109). Never assume exclusive access to shared resources — other MCP servers may be operating concurrently.
highJ5Tool Output Poisoning PatternsMCP01-prompt-injectionAML.T0054
[AST — J5] Catch block at L1467 interpolates error variable "err" into response. If the error originates from attacker-controlled input (e.g., malformed data), the error message becomes an injection vector into the AI's context.
Never include user input or LLM manipulation directives in error messages or tool responses. Use structured error codes.
highJ5Tool Output Poisoning PatternsMCP01-prompt-injectionAML.T0054
[AST — J5] Catch block at L1493 interpolates error variable "err" into response. If the error originates from attacker-controlled input (e.g., malformed data), the error message becomes an injection vector into the AI's context.
Never include user input or LLM manipulation directives in error messages or tool responses. Use structured error codes.
highC3Server-Side Request Forgery (SSRF)MCP04-data-exfiltrationAML.T0057
Pattern "new\s+URL\s*\(\s*(?:req|request|input|param|params|args|url|uri|href|link|target|destination|endpoint|host|address|resource|src|source)" matched in source_code: "new URL(url" (at position 30820)
Validate ALL user-supplied URLs before making HTTP requests:
1. Parse the URL and check the hostname against an explicit allowlist of permitted domains.
2. Block requests to RFC 1918 private ranges: 10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16.
3. Block loopback (127.0.0.0/8), link-local (169.254.0.0/16), and IPv6 equivalents.
4. Block file:// and other non-http(s) protocols explicitly.
5. Disable automatic redirect following, or re-validate each redirect destination.
6. In cloud environments: block requests to IMDS endpoints (169.254.169.254,
metadata.google.internal) at both the application AND network layer.
Example (Node.js): Use the `ssrf-req-filter` package or implement URL validation
against an allowlist before calling fetch/axios/got.
highD1Known CVEs in DependenciesMCP08-dependency-vuln
Dependency "@modelcontextprotocol/sdk@1.0.0" has known CVEs:
Update dependencies to versions that patch known CVEs. Run 'npm audit fix' or 'pip-audit' to identify and resolve vulnerable dependencies.
highJ5Tool Output Poisoning PatternsMCP01-prompt-injectionAML.T0054
[AST — J5] Catch block at L1394 interpolates error variable "err" into response. If the error originates from attacker-controlled input (e.g., malformed data), the error message becomes an injection vector into the AI's context.
Never include user input or LLM manipulation directives in error messages or tool responses. Use structured error codes.
Medium1
mediumK17Missing Timeout or Circuit BreakerMCP07-insecure-configAML.T0054
Pattern "(?:query|execute|find|select|aggregate)\s*\((?!.*(?:timeout|maxTimeMS|statement_timeout|deadline|cancel))" matched in source_code: "execute(" (at position 11791)
Add timeouts to ALL external calls: HTTP requests (30s), database queries (10s), subprocess execution (60s), and MCP tool calls (30s). Implement circuit breakers that open after N consecutive failures (e.g., opossum, cockatiel). Use AbortSignal for cancellable operations. Required by EU AI Act Art. 15 and OWASP ASI08.
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.