AI skills for LP teams
Real workflows we run inside FundFrame, shared as portable files. Paste a skill into Claude, Microsoft Copilot, or any other modern LLM and start running it in minutes.
The skill library
Tagged by FundFrame module so you can see where each one fits. Skills are portable recipes for Claude, Microsoft Copilot, or any other modern LLM.
How a skill works
A skill is a SKILL.md file that grounds an LLM in your workflow. The description tells the model when to load it, the body tells it what to do. Here is what that looks like in practice.
A single Markdown file
A skill is a SKILL.md file with a name and description in the frontmatter. The description tells the model when to load it. Everything else, scope, steps, rules, goes in the body.
Progressive disclosure
Only the name and description load at first. The full file loads when the task matches. Scripts and references load only when the workflow needs them. Many skills installed, minimal tokens spent.
Works everywhere
Drop a skill in ~/.claude/skills/ for personal use, in .claude/skills/ for your team via git, or bundle it with a plugin. Same file, different reach.
Bundle what the task needs
A skill can include scripts/ for deterministic code, references/ for detailed documentation, and assets/ for templates or sample files. Each loads only when the workflow calls for it.
Build one when you repeat yourself
If you keep giving the same instructions, pasting the same checklist, or your CLAUDE.md has grown into full procedures, that is a skill waiting to be extracted.
When the recipe is written down, the model stops guessing. You get the same workflow every time, from anyone on the team, in any tool that reads the file.
See how a skill works in practice--- name: pe-market-mapping description: Builds a continuously maintained map of the PE landscape for any geography or segment. Use for LP coverage and exposure work. --- # Scope Map the private equity landscape for a given geography or segment. Use only for LP coverage and exposure work, not portfolio diligence. # Workflow 1. Ask five scoping questions: use case, geography, strategy, size band, LP structure. 2. Run the discovery sweep. 3. Produce the five deliverables. # Decision rules - IF use case is "coverage trip" THEN require physical office in scope. - IF firm has multiple sub-strategies THEN list each qualifying fund line separately. # Guardrails - Do not infer fundraising status. - Do not output GPs without at least two corroborating sources.
Where to run them
The same recipe runs anywhere: as a first-class feature inside the FundFrame platform on your own fund data, or as a portable skill file you can upload to Claude or drop into Microsoft Copilot, Gemini, and other modern LLMs.
Built by LPs, for LPs. These reflect workflows we've automated inside the FundFrame platform.