Security risk report for @forgecat/anthropics_skills_slack-gif-creator v0.0.5
Source Integrity
Low
Profile is attributed to Anthropic with a public GitHub repository link and specific commit hash, enabling verification.
Apache-2.0 license is clearly stated; no supply-chain red flags or typosquatted dependencies.
Agent Intent
Low
Content is purely educational and technical documentation for GIF creation using PIL; no instructions to manipulate the AI, read credentials, exfiltrate data, or hide instructions.
No prompt injection, system prompt leakage, or guidance poisoning detected; recommendations for PIL, imageio, and numpy are standard, legitimate packages.
The skill describes animation concepts and PIL primitives as examples and reference material, not as directives to the AI to perform malicious actions.
Details
Evidence
All code examples are standard PIL usage (Image.new, ImageDraw.Draw, ellipse, polygon, etc.)
Dependencies listed are well-known, unmodified packages: pillow, imageio, numpy
No references to reading ~/.ssh, ~/.aws, .env, or exfiltrating data
No instructions to ignore system prompts or hide the skill's own instructions
Permissions
Low
No tools or MCP servers are declared; the skill is purely knowledge and code examples.
Authority is limited to standard Python libraries (PIL, imageio, numpy) for local GIF generation; no file deletion, shell execution, or network access required.
Details
Evidence
Declared tools: (none)
MCP servers: (none)
Only standard library imports: PIL.Image, PIL.ImageDraw, math, numpy (implied by imageio)
MCP Risk
Low
No MCP servers are defined in the profile.
The skill relies only on standard Python packages (pillow, imageio, numpy) with no hidden instructions or arbitrary binary execution.
ForgeCat