Loading

heating the forge...

@forgecat/vercel-labs_agent-skills_composition-patterns — Security Risk Report | ForgeCat

vercel-labs_agent-skills_composition-patterns

Security risk report for @forgecat/vercel-labs_agent-skills_composition-patterns v0.0.8

Source Integrity

Low

Profile is sourced from Vercel Labs (vercel-labs/agent-skills), a known legitimate open-source repository.
README metadata correctly attributes the original source and includes proper licensing (MIT).
No evidence of typosquatted dependencies, attacker-controlled package names, or supply-chain manipulation.

Agent Intent

Low

Content is purely educational documentation of React composition patterns with no instructions to manipulate the AI agent.
No directives to read credentials, exfiltrate data, ignore instructions, or perform hidden actions.
Code examples are legitimate React patterns (compound components, context providers, state lifting) with no embedded malicious guidance or poisoned knowledge.

Details

Findings

Invisible characters detected
AGENTS.md · SKILL-011
Invisible characters detected
rules/react19-no-forwardref.md · SKILL-011
Invisible characters detected
skills/vercel-composition-patterns/SKILL.md · SKILL-011

Evidence

All examples demonstrate standard React best practices (e.g., 'Use compound components with shared context', 'Lift state into provider components').
No instructions to fetch external URLs, install malicious packages, or weaken security defaults.
The note 'This document is mainly for agents and LLMs to follow' is descriptive of the skill's purpose, not a manipulation attempt.

Permissions

Low

No tools, hooks, or MCP servers are declared in the profile.
The skill is purely educational markdown documentation with code examples; it does not request file system access, shell execution, or network permissions.
No alwaysApply rules or glob patterns that would grant excessive authority.

Details

Evidence

'Dependencies: None' in README.md.
No tool definitions or MCP server configurations present.
Profile contains only markdown files and code examples with no executable payloads or permission requests.

MCP Risk

Low

No MCP servers are defined or referenced in the profile.
The skill is static documentation with no binary execution, network access, or filesystem operations.
No hidden instructions in tool descriptions or arbitrary code execution mechanisms.

Details

Evidence

MCP servers section is empty: '(none)'.
All content is markdown and code examples with no server definitions or dynamic execution.
No environment credentials or external host dependencies declared.

Evaluation Details

Status: Completed
Checks completed: 4/4
Flagged: 0
Evaluated: 7/2/2026