My `bcond' macro, which works closely with my `nlet' macro, attempts to solve the same problem. `nlet' subsumes `multiple-value-bind':
(nlet ((a b (foo)))
(bar a b))
Coming up with a good syntax for `bcond' wasn't easy, and I'm not sure what you'll think of my solution, but here it is. It's like `cond', except that if the test subform is an `nlet' form, the scope of the bindings it establishes is magically extended to included the entire clause. E.g.:
(bcond ((nlet ((a b (foo)))
(test a b))
(do-something a b))
... more clauses ...)
The expansion is careful not to extend the scope of the bound variables over the following clauses.
These macros are in my "misc-extensions" project on
common-lisp.net, included in LibCL and Quicklisp. As they are upward-compatible with `cl:let' and `cl:cond', the package provides the same macros under the names `let' and `cond', which can be shadowing-imported if desired (I use them this way).
`nlet' also subsumes and generalizes `let*'.
BTW I wrote these 30 years ago, in ZetaLisp. I've always made heavy use of multiple values in my style, and `multiple-value-bind' is way too inconvenient to use -- the name is too long and it only takes one binding clause at a time.
-- Scott