Security risk report for @forgecat/s1dashu_ip-as-logo-skill v0.1.3
Source Integrity
Low
Profile references a legitimate public GitHub repository (s1dashu/ip-as-logo-skill) with a verifiable commit hash.
No instructions to fetch untrusted external URLs, install attacker-controlled dependencies, or pull from typosquatted packages.
Agent Intent
Low
Content describes a legitimate image-generation skill with no instructions to manipulate the AI's behavior, ignore system instructions, or exfiltrate data.
No guidance poisoning: the skill recommends standard image-generation practices and well-known platforms (Codex, Cursor, Claude Code) without malicious steering.
No system prompt leakage or hidden instructions; the markdown documents a straightforward workflow for generating mascot images.
Details
Evidence
Workflow steps are transparent and user-centric: 'Parse the request', 'inspect relevant read-only context', 'present three concise directions', 'generate six independent candidates'.
Complexity budget and shape language are legitimate design constraints, not attack vectors.
No references to reading ~/.ssh, ~/.aws, .env, or exfiltrating via curl/POST.
Permissions
Low
No tools or hooks are declared in the profile; the skill is purely instructional guidance for an image-generation capability.
The skill requests only read-only context inspection (README, docs, metadata) and image generation—both aligned with its stated purpose.
No file_write, file_delete, shell, or excessive web_fetch permissions are requested.
Details
Evidence
Workflow step 2: 'inspect relevant read-only context before asking questions. Prefer the README, product docs, package or app metadata...'
No alwaysApply rules, glob patterns, or high-risk tool categories are present.
MCP Risk
Low
No MCP servers are defined in the profile.
The skill depends on a configured image-generation capability (Codex ImageGen, or a generic image generator) but does not define or execute arbitrary binaries.
Details
Evidence
Profile declares: '(none)' for MCP servers.
Workflow step 8 states: 'Determine the available image-generation path before promising output... if none is available, ask the user whether they can provide or enable one. Do not fabricate generated results.'
ForgeCat