MIT license and clear attribution to original author; no supply-chain red flags.
Agent Intent
Low
Content describes a documentation-lookup utility with no instructions to manipulate the AI, exfiltrate data, or bypass security controls.
Agent instructions focus on legitimate library resolution and documentation fetching; no prompt injection, role hijacking, or guidance poisoning detected.
No directives to read credentials, install malicious dependencies, weaken security defaults, or leak system prompts.
Details
Evidence
Agent task: 'fetch the relevant documentation and return a concise, actionable answer with code examples'
Skill activation: 'use Context7 to fetch current documentation instead of relying on training data'
Guidelines emphasize accuracy and official sources, not deception or attacker goals.
Permissions
Low
No explicit tool declarations in the profile; authority is limited to calling two MCP tools (resolve-library-id, query-docs) for documentation lookup.
MCP tools are scoped to their stated purpose: resolving library identifiers and fetching documentation from Context7.
No file system access, shell execution, or credential reading requested.
Details
Evidence
Only tools mentioned: 'resolve-library-id' and 'query-docs' for documentation retrieval.
No alwaysApply rules, glob patterns, or high-risk categories (shell, file_write, file_delete).
Skill description limits activation to documentation and code-example scenarios.
MCP Risk
MCP-000Medium
The MCP server points to an external third-party host (`https://mcp.context7.com/mcp`) operated by Upstash, meaning all documentation queries are routed through an external service outside the user's control.
No hidden instructions were found in tool descriptions, and the binary (HTTP MCP) is a known pattern; however, reliance on an external host introduces a dependency on that host's integrity and availability.
No credentials are required and no unrestricted filesystem or network access is granted, keeping overall risk moderate rather than high.
ForgeCat