| name | brain-start |
| description | Front door for plain-English Company Brain requests when the user's role or next step is unclear. Use when the user says I want to set up my company brain, help me start, I want to use the brain, what should I do next, create a company brain, operate the brain, or add knowledge but the right role is not clear. |
Brain Start
Be the front door. Hide the harness internals unless the user asks. Do not start by listing CLIs.
User Promise
The user should feel like this:
I say what I want.
The agent asks a few simple questions.
The agent runs the harness safely in the background.
I review the next action.
Workflow
- Resolve whether a brain root already exists.
- Ask one question: "Are you setting this up, operating it, or contributing knowledge?"
- Route by role:
- setup/owner -> use
brain-owner
- daily checks/maintenance -> use
brain-operator
- teammate contribution -> use
brain-contribute
- If the user is unsure, default to owner setup.
- Explain only the next step, not the whole system.
- Run lower-level skills or CLIs internally as needed.
- End with one clear next action.
Runtime Assumption
This skill is normally loaded from an installed plugin bundle. The bundle
already includes skills/, bin/, and references/.
- Do not tell a normal user to download individual skills.
- Do not tell a normal user to clone the harness repo unless they are doing
local development or auditing the package.
- When a script is needed, resolve it from the installed plugin root's
bin/
directory.
- Treat the company brain root as customer data, separate from the installed
harness plugin.
Role Names
Use these public names:
- Brain Owner: approves policy, sources, restricted access, and launch readiness.
- Brain Operator: runs daily checks, reviews queue, and keeps the brain healthy.
- Team Member: contributes role knowledge and flags stale or private material.
Do not lead with internal role labels, source-owner, registry, or CLI language unless the user is technical or asks.
First Questions
Ask these progressively, one at a time:
- "Are you creating a new company brain or using an existing one?"
- "Where should the brain live?"
- "Are you the Brain Owner, Brain Operator, or a Team Member?"
- "Do you want the 20-minute starter setup or the team pilot setup?"
- For setup: "Do you want the governed review model or the simpler team drop-zone model?"
Background Actions
When needed, run:
python3 <plugin-root>/bin/brain-setup.py ...
python3 <plugin-root>/bin/connections-check.py ...
python3 <plugin-root>/bin/source-registry-check.py ...
python3 <plugin-root>/bin/brain-health.py ...
python3 <plugin-root>/bin/brain-lint.py ...
python3 <plugin-root>/bin/brain-schedule.py ...
python3 <plugin-root>/bin/add-to-brain-digest.py ...
Show results as plain-language status:
- Ready
- Needs review
- Blocked
- Next action
Guardrails
- Setup and schedule commands preview first unless the user explicitly asks to create/update files.
- Do not populate company facts during setup.
- Use
--operating-profile simple-team only when the owner wants a shared drop zone and digest automation.
- Do not connect personal accounts as company sources.
- Do not promote governed-profile, sensitive, or unclear notes without approval.
- Do not inspect restricted folders.