The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill documentation in cli-contract.md describes accessing and using sensitive local credential stores located at ~/.paperclip/auth.json and ~/.paperclip/context.json, as well as managing the PAPERCLIP_API_KEY environment variable.
[EXTERNAL_DOWNLOADS]: The cli-contract.md file contains instructions to install the paperclipai package globally via npm install -g.
[REMOTE_CODE_EXECUTION]: The skill utilizes npx -y @bbengamin/paperclip-mcp-server to dynamically execute code from a remote registry at runtime.
[COMMAND_EXECUTION]: The skill relies on shell commands including curl, npm, and npx for system interaction. Automated scanners identified suspicious HTTP request patterns (MD:HttpRequest-inf) in cli-contract.md related to the construction of authenticated API calls.
[DATA_EXFILTRATION]: A URL referenced in the documentation (https://paperclip.right.link/RL/wiki/page/wiki/sources/rl-30-day-validation-plan.md) has been blacklisted by security scanners. The skill is designed to fetch content from such external sources via the paperclip-wiki-fetch tool. This, combined with access to local authentication secrets and the capability to perform network requests via curl to user-defined API endpoints, presents a significant risk of credential exfiltration.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface through its processing of untrusted external wiki data.
Ingestion points: SKILL.md (via content retrieved by the paperclip-wiki-fetch tool).
Boundary markers: Absent.
Capability inventory: Command execution via the paperclipai CLI and network/file mutations via Paperclip MCP tools.
Sanitization: Absent.

creator-plan-work