Claude Skills that turn your AI agent into an expert business operator

Anthropic user adatarwa released 11 installable Claude Skills that turn the AI agent into domain-specific business operators, including sales, finance, HR, and legal roles. The skills encode triggering logic, guardrails, and output discipline, and are available on GitHub for use with Claude Code, Claude Cowork, or Claude.ai projects.

11 installable Claude Skills that turn your AI agent into an expert business operator. Drop these into Claude Code, Claude Cowork, or any Claude.ai project, and your AI stops giving generic advice and starts behaving like an experienced Sales lead, CFO, HR director, or General Counsel — with the specific judgment calls, output formats, and guardrails real operators actually use. This is the companion project to ai-for-operators https://github.com/adatarwa/ai-for-operators , which is the human-readable prompt library. This repo packages that same expertise as Agent Skills — so instead of copy-pasting prompts, you install a skill once and Claude automatically applies the right behavior every time. A Claude Skill is a folder containing a SKILL.md file that teaches Claude how to handle a specific domain of work — when to apply specialized behavior, what output format to use, and what guardrails to respect. Skills load automatically when relevant to what you're asking, with no special syntax or commands needed. Read more: Anthropic's Skills documentation https://docs.claude.com | Skill | What it does | |---|---| sales-operator | operations-operator finance-operator hr-operator procurement-operator marketing-operator customer-success-operator legal-operator customer-support-operator product-operator founder-operator Every skill ships with a references/prompt-library.md containing the full set of tested prompt patterns it's built from — Claude consults these automatically when a request matches a specific scenario. Every skill encodes three things a plain prompt can't: Triggering logic — the skill activates automatically on relevant requests, even casually phrased ones "this customer seems quiet lately" triggers customer-success-operator without the user saying "churn risk" . Domain guardrails — legal-operator and hr-operator explicitly require flagging when something needs qualified professional review. finance-operator explicitly refuses to invent financial figures. Output discipline — word limits, required structure e.g. always show scoring reasoning, not just a ranking , and defaults that match how real operators actually communicate. git clone https://github.com/adatarwa/operator-skills.git Claude Code uses the folder name not the name: field inside SKILL.md as the skill identifier. Copy the skill folder s you want directly into ~/.claude/skills/ : cp -r operator-skills/skills/sales ~/.claude/skills/sales-operator cp -r operator-skills/skills/finance ~/.claude/skills/finance-operator ...or copy all 11 at once, renaming each to match its name: field: for skill in sales operations finance hr procurement marketing customer-success legal customer-support product founder; do cp -r "operator-skills/skills/$skill" "~/.claude/skills/${skill}-operator" done Restart Claude Code skills load at session startup . Verify installation by typing /skills — you should see all installed skills listed. - Download the skill folder s you want from skills/ - In your Claude.ai Project settings, upload the SKILL.md and references/ folder as project knowledge - Claude will apply the skill's behavior within that project automatically Type /skills in Claude Code to confirm the skill appears in your installed list. Then test triggering with something casual — not the skill's exact name: "This customer hasn't logged in for 3 weeks, should I be worried?" If customer-success-operator is installed correctly, Claude responds with a structured risk assessment risk level, likely cause, 14-day action rather than generic advice. Skills compose naturally. A founder working on a fundraise might have founder-operator and finance-operator both installed — Claude will draw on both when the request touches investor narrative and financial variance explanation in the same conversation. operator-skills/ ├── skills/ │ ├── sales/ │ │ ├── SKILL.md │ │ └── references/prompt-library.md │ ├── operations/ │ │ ├── SKILL.md │ │ └── references/prompt-library.md │ ├── finance/ │ ├── hr/ │ ├── procurement/ │ ├── marketing/ │ ├── customer-success/ │ ├── legal/ │ ├── customer-support/ │ ├── product/ │ └── founder/ ├── CONTRIBUTING.md ├── LICENSE └── README.md ai-for-operators https://github.com/adatarwa/ai-for-operators — the human-readable version of this same expertise. Read it in a browser, copy a prompt, paste it manually into any AI tool. Use that repo if you're not running Claude Code/Cowork; use this repo if you are. Found a gap in a skill's behavior? Want to add a 12th skill e.g. engineering management, recruiting ? See CONTRIBUTING.md /adatarwa/operator-skills/blob/main/CONTRIBUTING.md . MIT — use it, fork it, build on it. Built for operators who run their work through Claude.