Back to skill

Security audit

Reach

Security checks for vulnerabilities and agentic risk

Overview

This skill broadly matches its advertised web-agent purpose, but its defaults expose sensitive sessions, inbound webhooks, and payment/signing actions without enough safeguards.

Use this only in an isolated environment with low-privilege service accounts and a limited wallet. Do not point it at high-value accounts, shared folders, or synced repositories unless credential storage is hardened. Keep the webhook server bound to localhost or behind authenticated ingress, require explicit review before email/sign/pay actions, and set strict payment limits for x402 flows.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (4)

T09 · Insecure Skill Coding Practices

Error
Location
src/utils/webhook-server.js:24
Finding

Unauthenticated Network-Exposed Webhook Permits Forged Email and Agent Events

Content
View full analysis
{ receiveEmail(payload); }); } } } ``` ```javascript this.server.listen(this.port, this.host, () => { console.log(`[webhook] Server listening on ${this.host}:${this.port}`); console.log(`[webhook] ${this.handlers.size} handlers registered`); resolve(this); }); ``` ```javascript // Check registered handlers const handler = this.handlers.get(path); if (handler) { try { handler(payload, req.headers, req); res.writeHead(200, { 'Content-Type': 'application/json' }); res.end(JSON.stringify({ received: true, path })); return; } catch (e) { console.log(`[webhook] Handler error for ${path}: ${e.message}`); res.writeHead(500, { 'Content-Type': 'application/json' }); res.end(JSON.stringify({ error: e.message })); return; } } // Check observe webhooks (registered via observe(..., { method: 'webhook' })) if (globalThis.__reachWebhooks) { const observeHandler = globalThis.__reachWebhooks.get(path); if (observeHandler && observeHandler.callback) { try { observeHandler.callback({ type: 'webhook', target: observeHandler.target, path, payload, headers: req.headers, timestamp: new Date().toISOString(), }); res.writeHead(200, { 'Content-Type': 'application/json' }); res.end(JSON.stringify({ received: true, path, observer: observeHandler.id })) ...[truncated 2323 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
src/primitives/authenticate.js:205
Finding

Session Cookies and API Keys Are Persisted as Unprotected Plaintext Files

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
src/primitives/fetch.js:50
Finding

Unrestricted Agent-Controlled URL Fetching Enables Server-Side Request Forgery

Content
View full analysis
50) { return result; } console.log('[fetch] HTTP content too short, trying browser'); } catch (e) { console.log(`[fetch] HTTP failed: ${e.message}, trying browser`); } return await fetchWithBrowser(url, format, session); } ``` ```javascript async function fetchWithHTTP(url, format, headers = {}) { const response = await nodeFetch(url, { headers: { 'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/123.0.0.0 Safari/537.36', 'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8', ...headers, }, timeout: 15000, }); ``` ```javascript async function fetchWithBrowser(url, format, session) { const domain = pool.getDomain(url); const page = awa ...[truncated 2436 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
src/primitives/pay.js:35
Finding

x402 Authorization Ignores the Caller-Supplied Amount and Trusts Unbounded Server Terms

Content
View full analysis
parseFloat(maxAmount)) { return { success: false, type: 'x402', error: `Payment amount ${paymentDetails.amount} exceeds max ${maxAmount}`, paymentDetails, }; } console.log(`[pay] x402 requires: ${pa ...[truncated 3293 chars]
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (110)

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · README.md (reported line 541)May include surrounding context.

md
| `RPC_URL` | No | RPC endpoint (default: Base mainnet) |
| `RESEND_API_KEY` | For email | Resend API key (send from @exoagent.xyz or @mfer.one) |
| `CAPSOLVER_API_KEY` | For CAPTCHA | CapSolver API key |
| `GITHUB_TOKEN` | For GitHub API | GitHub personal access token |

## License

Tp4

High
Category
MCP Tool Poisoning
Confidence
88% confidence
Finding

Starting an inbound webhook/listener service materially changes the trust and exposure model of the skill. If users expect outbound-only automation, they may not harden the host or network perimeter appropriately, increasing the risk of unauthorized requests or abuse of trigger infrastructure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Starting an inbound webhook/listener service materially changes the trust and exposure model of the skill. If users expect outbound-only automation, they may not harden the host or network perimeter appropriately, increasing the risk of unauthorized requests or abuse of trigger infrastructure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

Starting an inbound webhook/listener service materially changes the trust and exposure model of the skill. If users expect outbound-only automation, they may not harden the host or network perimeter appropriately, increasing the risk of unauthorized requests or abuse of trigger infrastructure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

Starting an inbound webhook/listener service materially changes the trust and exposure model of the skill. If users expect outbound-only automation, they may not harden the host or network perimeter appropriately, increasing the risk of unauthorized requests or abuse of trigger infrastructure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Starting an inbound webhook/listener service materially changes the trust and exposure model of the skill. If users expect outbound-only automation, they may not harden the host or network perimeter appropriately, increasing the risk of unauthorized requests or abuse of trigger infrastructure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

Starting an inbound webhook/listener service materially changes the trust and exposure model of the skill. If users expect outbound-only automation, they may not harden the host or network perimeter appropriately, increasing the risk of unauthorized requests or abuse of trigger infrastructure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Starting an inbound webhook/listener service materially changes the trust and exposure model of the skill. If users expect outbound-only automation, they may not harden the host or network perimeter appropriately, increasing the risk of unauthorized requests or abuse of trigger infrastructure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Starting an inbound webhook/listener service materially changes the trust and exposure model of the skill. If users expect outbound-only automation, they may not harden the host or network perimeter appropriately, increasing the risk of unauthorized requests or abuse of trigger infrastructure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
89% confidence
Finding

Starting an inbound webhook/listener service materially changes the trust and exposure model of the skill. If users expect outbound-only automation, they may not harden the host or network perimeter appropriately, increasing the risk of unauthorized requests or abuse of trigger infrastructure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

Starting an inbound webhook/listener service materially changes the trust and exposure model of the skill. If users expect outbound-only automation, they may not harden the host or network perimeter appropriately, increasing the risk of unauthorized requests or abuse of trigger infrastructure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Starting an inbound webhook/listener service materially changes the trust and exposure model of the skill. If users expect outbound-only automation, they may not harden the host or network perimeter appropriately, increasing the risk of unauthorized requests or abuse of trigger infrastructure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Starting an inbound webhook/listener service materially changes the trust and exposure model of the skill. If users expect outbound-only automation, they may not harden the host or network perimeter appropriately, increasing the risk of unauthorized requests or abuse of trigger infrastructure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Starting an inbound webhook/listener service materially changes the trust and exposure model of the skill. If users expect outbound-only automation, they may not harden the host or network perimeter appropriately, increasing the risk of unauthorized requests or abuse of trigger infrastructure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Starting an inbound webhook/listener service materially changes the trust and exposure model of the skill. If users expect outbound-only automation, they may not harden the host or network perimeter appropriately, increasing the risk of unauthorized requests or abuse of trigger infrastructure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Starting an inbound webhook/listener service materially changes the trust and exposure model of the skill. If users expect outbound-only automation, they may not harden the host or network perimeter appropriately, increasing the risk of unauthorized requests or abuse of trigger infrastructure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

Starting an inbound webhook/listener service materially changes the trust and exposure model of the skill. If users expect outbound-only automation, they may not harden the host or network perimeter appropriately, increasing the risk of unauthorized requests or abuse of trigger infrastructure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Starting an inbound webhook/listener service materially changes the trust and exposure model of the skill. If users expect outbound-only automation, they may not harden the host or network perimeter appropriately, increasing the risk of unauthorized requests or abuse of trigger infrastructure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

Starting an inbound webhook/listener service materially changes the trust and exposure model of the skill. If users expect outbound-only automation, they may not harden the host or network perimeter appropriately, increasing the risk of unauthorized requests or abuse of trigger infrastructure.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill advertises high-impact capabilities including logging into services, signing transactions, payments, and sending email, but it provides no prominent user warnings or consent model. In agent environments, this can normalize dangerous actions and lead users to expose credentials or authorize side-effecting operations without understanding the consequences.

Content

No source excerpt is available for this finding.

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 103)May include surrounding context.

md
node src/cli.js fetch https://example.com

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 104)May include surrounding context.

md
node src/cli.js fetch https://example.com

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 105)May include surrounding context.

md
node src/cli.js fetch https://example.com

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 106)May include surrounding context.

md
node src/cli.js fetch https://example.com

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 107)May include surrounding context.

md
node src/cli.js fetch https://example.com