Wouldwork is a classical AI planning environment implemented in SBCL Common Lisp, PDDL-flavored, REPL-friendly.
Users describe a world (objects, relations, actions, constraints, goals, ...), and Wouldwork searches for plans that reach your goal. It enables users to solve planning and constraint satisfaction problems without extensive programming experience. The typical workflow is to create a problem spec → stage the problem → set search parameters → solve the problem → review results. The environment is highly interactive to support experimentation, exploration, debugging, and analysis.
Wouldwork supports most “classic planner” features, including an accessible Lisp-native foundation:
- Planning and Constraint Satisfaction Problems
- Objects, types, relations (dynamic + static)
- Actions with preconditions/effects
- Goals, constraints, and derived relations
- Functions (including on-the-fly and recursive computations)
- Exogenous events (things that happen independently of the agent)
- Temporal planning (action schedules / timestamps)
- Search control: depth cutoffs, tree vs graph search, randomization, branching
- Solution modes: first solution, shortest plan, min-time, min/max objective, all solutions, ...
- Optional parallel search (using SBCL & Bordeau Threads primitives)
- Diagnostics & debugging hooks (including step-through search)
For a gentle narrative intro, see the Medium article: https://medium.com/@davypough/traditional-ai-problem-solving-with-wouldwork-fcb0c4a71226
GitHub: https://github.com/davypough/wouldwork
See the User Manual for tutorials and explanations on how to use the program
Questions/Comments/Suggestions: [email protected]
- First install SBCL and Quicklisp
- Start SBCL and at the prompt enter (progn (ql:quickload :wouldwork) (in-package :ww))
- Then at the prompt enter (help) or (test) to run Wouldwork on all the test problems to verify correctness
Wouldwork keeps your working context in a small file (vals.lisp) inside the system directory.
That means when you restart SBCL and Wouldwork, it will restore:
- the problem you were last working on
- all your parameter choices for that problem (algorithm, search settings, debug level, ...).
Accepted solve-subgoal checkpoints are deliberately not stored in vals.lisp. To carry
an incomplete milestone chain across a restart, use
(export-subgoal-progress "name.txt") before leaving. After restarting, freshly stage the
same problem and use (import-subgoal-progress "name.txt"). The text file lists the goals,
settings, actions, and symbolic checkpoint facts for inspection. Import replays every action
and reruns the problem's ordinary and technology-specific validators before restoring the
chain; it never trusts a serialized in-memory planning state.
A milestone that cannot be reached rewinds to the preceding milestone and looks for another
checkpoint, but only when its search actually explored the whole space. When the depth cutoff
cut off part of that space instead, solve-subgoal reports the truncation, rejects no
checkpoint, and returns with the accepted chain unchanged -- raise *depth-cutoff* and
reissue the same solve-subgoal.
This is deliberate: planning development is exploratory, and it avoids re-entering all the search control parameters every new session.
There’s also a small but important design choice:
- when you call (stage problem-name), Wouldwork loads that problem spec with fresh defaults
- when you call (refresh), Wouldwork reloads the current problem spec, but avoids clobbering any parameter values you just set at the REPL.
If you ever want to start over, as if beginning a new session, enter (ww-reset) from the REPL.
Wouldwork includes a lot of features for handling/specifying complex problems (see the User Manual). Start with thoroughly understanding and solving a simple problem like problem-blocks3.lisp first, before tackling more complex problems. Selecting a good problem representation and solution strategy may determine whether Wouldwork can find a solution. Note that modern LLMs like Claude or chatGPT are quite good at drafting a problem specification, if given an accurate problem description along with the user manual.