System prompt · Context file · Run prompts · First-run test
A chatbot waits for a clever prompt. An agent starts with a job. This free starter pack gives five business agents the same structure Brobots uses: a system prompt that defines the work, a context file that teaches the business, repeatable run prompts, and a first-run test that exposes bad output before it touches real work.
Download the complete pack (.txt)
No email gate. Copy one kit at a time below, or take the whole pack.
Turn scattered updates into a five-minute decision brief.
Open kit ->Agent 02Find out whether AI can find, understand, trust, and recommend your business.
Open kit ->Agent 03Draft posts, emails, and pages in your voice from real source material.
Open kit ->Agent 04Turn a messy service into a clear package buyers can understand and act on.
Open kit ->Agent 05Score inquiries, show the evidence, name the missing questions, and draft the reply.
Open kit ->Replace the brackets with real business facts. Write UNKNOWN instead of guessing. This one file becomes the source of truth for all five agents.
Paste its system prompt into a Claude Project, ChatGPT Project, Custom GPT, Gem, or new chat. Add the Universal Context plus the agent addendum.
Run the included first-run test. If the agent invents facts, misses the trap, or ignores the format, fix the context before giving it real work.
This is the part most prompt packs skip. The context file teaches an agent what you sell, who it is for, what proof it may use, how you sound, and what it must never invent.
Turn scattered updates into a five-minute decision brief.
Best for owners juggling calendar, inbox, customers, and projects.
Run cadence: Weekdays before work. Monday planning version, Friday review version.
Find out whether AI can find, understand, trust, and recommend your business.
Best for any business that wants AI search to send buyers, not just traffic.
Run cadence: Monthly, after major site changes, and whenever competitors change.
Draft posts, emails, and pages in your voice from real source material.
Best for owners who know the business but hate starting from a blank page.
Run cadence: Whenever real source material exists. Weekly batching works best.
Turn a messy service into a clear package buyers can understand and act on.
Best for services people keep asking you to explain twice.
Run cadence: On demand when services or prices change. Review live offers monthly.
Score inquiries, show the evidence, name the missing questions, and draft the reply.
Best for businesses losing time to unqualified or incomplete inquiries.
Run cadence: As leads arrive, with a daily or weekly pipeline review.
An agent can only check a calendar, inbox, website, CRM, or analytics source when your AI setup actually gives it access. Every kit tells it to name missing sources instead of pretending.
Some AI plans support scheduled tasks and connected tools. Some do not. Run each agent manually until its output passes, then schedule only what your platform can really run.
Customer messages, prices, claims, bookings, quotes, posts, and commitments stay under human approval. The agent drafts and analyzes. You decide.
Want Brobots to build one for your business?
The free pack shows the machine. A custom Brobots agent is built around one real workflow, your actual business context, explicit handoffs, and a human approval step. Custom agent: $3,000. DIY agent access: $99/mo. See the full menu.