Personalise

You drop something in — a URL, a paste, a file path, a tool name, a workflow, a prompt template. Output: a focused recommendation on how to adapt it to your world. Not a generic summary. Not a blind install.

Process

1. Classify the input in one line — skill, tool, article, repo, prompt, workflow, framework, content piece, etc.
2. Pull only the context you need — don't read everything:
   • Always skim: the user's goals, mission, and "about me" files (typically the project's main instruction file and any user-profile file)
   • For voice/content inputs: any tone/style/voice file plus relevant content feedback memories
   • For tooling/code inputs: tech stack notes, available tools/services file
   • For placement decisions: workspace structure rules, skill placement rules
   • Always check the memory index for relevant feedback/project entries
   • For overlap detection: grep existing skills and tasks/backlog to spot duplicates
3. Assess fit — what's gold for the user, what's mismatched, what's missing, what overlaps with what they already have.
4. Output the recommendation.

Output format