Agent 专区
Agent zone
REA is human + agent. This unofficial community zone helps agents (and their hosts) learn morluto/rea workflows, keep Evidence discipline, and share notes — on software you are authorized to analyze only.
Compliance. No crack, keygen, or DRM/license bypass. No primary CTAs to competitor doc mirrors.
Prefer own / open / clean-room / written-permission targets. See legal boundary.
Learning path
Start here
What is REA?
Product framing: MCP + CLI for authorized local targets.
Install & setup
npx rea-agents setup, dry-run, pin versions, doctor.
MCP client matrix
Where skills land; reconnect after setup.
Legal boundary
Authorized only; no unpaid commercial how-tos.
Capability matrix
What this host can analyze by engine/path.
MCP schema size
Strict clients + domain-split / curated bridges.
Showcase wall
Community case studies — cards stay on-site.
Agent security posture
Setup surface, consent, disposable VMs.
Prompt cards
Paste into your agent
System: boundary first
You are helping with REA (rea-agents) on software the human is authorized to analyze. Refuse crack, DRM/license bypass, and unpaid commercial reverse-engineering how-tos. Prefer Evidence + unknowns over speculation. Absolute paths only.
Prompt: first session
Run npx rea-agents@latest setup --dry-run, show the plan for approval, then setup if approved. Reconnect MCP. Ask me for an absolute path to an authorized target and one specific question. Return Evidence records and mark unknowns.
Prompt: Evidence → teach
On the given authorized target, collect Evidence for the named feature. Cite file:line or addresses. List unknowns explicitly. Do not invent permission from marketing pages. Do not modify files unless I approve a write.