The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documents the installation and usage of the paperclipai CLI and the @bbengamin/paperclip-mcp-server via npm and npx. These are legitimate vendor-controlled resources (authored by bbengamin) essential for the skill's operation within the Paperclip ecosystem.
[COMMAND_EXECUTION]: The skill extensively uses shell command execution via the paperclipai CLI to orchestrate Paperclip-native records such as issues, goals, and projects, as well as managing wiki pages.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection when processing untrusted external data from Paperclip wiki pages and issue comments.
Ingestion points: Untrusted data enters the agent context through paperclip-wiki-fetch (reading wiki pages) and paperclipai issue get (reading comments).
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions when interpolating fetched data into proposed issue bodies.
Capability inventory: The skill has the capability to create/update issues via paperclipai issue create and modify wiki content via paperclip-wiki-manage.
Sanitization: No programmatic sanitization is defined; however, the risk is mitigated by a mandatory 'Approval Boundary' rule requiring human operator confirmation before any mutations are executed.

outbound-plan-work