Back to skill

Security audit

company-research

Security checks for vulnerabilities and agentic risk

Overview

The skill is a read-only company research helper that uses disclosed ReplyNodes web services to gather public company evidence.

Before installing, understand that the skill sends requested company domains and selected public URLs to ReplyNodes-operated services. Use a reviewed or pinned install source where possible, and store any optional REPLYNODES_API_KEY only in your agent's secret store, not in chat or files.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

Lines L021-L022 state the skill is read-only and 'needs no backend service,' which implies local or direct web-only operation. However, the file repeatedly directs the host to call ReplyNodes-operated network services such as md.replynodes.com, brand.replynodes.com, and optionally mcp.replynodes.com (for example L049-L053, L288-L293, L499-L506), so the stated operating model does not match the actual implementation described in the same file.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
80% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 336)May include surrounding context.

md
plus a `Retry-After` header (seconds to the next UTC midnight). Record the `429`
in `coverage_limits` and offer the user the existing free-account continuation at
<https://docs.replynodes.com/docs/auth> — an existing authenticated free account
has 500 credits. Never ask the user to paste an API key into chat. If the shared
Redis-backed quota or cache is unavailable the route fails closed with HTTP
`503 degraded`.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
82% confidence
Finding

Lines L319-L325 say there is no workflow route and that the composition uses 'free Markdown and Brand first, then the Tier B primitive routes when needed.' Earlier, L281-L286 instructs the agent to resolve company names 'only from the free keyless GET /v1/web/search route,' which is still a top-level API route used as part of the workflow. While this may intend to distinguish 'workflow route' from 'primitive route,' the wording is contradictory enough to misstate what routes are actually used.

Content

No source excerpt is available for this finding.