The Scope Boundary Definer

My agent's scope keeps expanding. Help me draw a boundary.

Original purpose: [DESCRIBE]
What people now want it to do: [LIST THE REQUESTS]
What it currently does: [DESCRIBE]

Produce:
1. A single sentence stating what this agent is for
2. Requests that fit that sentence
3. Requests that do not fit, with what should handle them instead
4. Requests that fit but should be deferred, with why
5. The cost of saying yes to each out-of-scope request: complexity
   added, failure modes introduced, evaluation burden created
6. The boundary statement I should put in the system prompt so the
   agent itself declines out-of-scope work
7. The user-facing explanation of what this agent does and does not
   do

Be firm. An agent that does eight things badly is worse than one
that does two things reliably, and the eight-thing version is
harder to evaluate, harder to debug, and harder to trust.