delete_rule_base
Delete a saved rule base by name.
LABSCO SUMMARY
Constraint and rule questions get an answer that was derived, not generated.
The model writes the facts and rules; the solver decides what follows from them, which is the part language models are worst at and Prolog is built for. The rule-base store is what makes it more than a scratchpad — encode a domain once, then ask questions against it for as long as it holds. trace is there when the answer is surprising and you need to see the derivation rather than take it.
What it is
A Prolog execution server with a named, reusable rule-base store behind it
What you get
- Execution of Prolog facts and rules against a query, with max_results capping the answers and trace exposing how they were derived (execute_prolog).
- CLP(FD) constraints, negation-as-failure and standard SWI-Prolog features, rather than a reduced subset.
- Named rule bases saved once and reused across execute_prolog calls, so stable knowledge such as chess piece_moves rules does not travel in every prompt (save_rule_base, get_rule_base).
- A listing of saved rule bases with the description and tags read from each file's leading comments, sorted by name (list_rule_bases), and removal by name (delete_rule_base).
Requirements
SWI-Prolog 9.0 or newer, installed and on PATH — the server runs your code on it. No account and no key: you supply the Prolog code and the query, and saved rule bases live with the server, addressed by name.
Setup effort
One command — pip install prolog-reasoner
USE CASES
01
Checking whether a set of rules really implies the conclusion
The facts, the rules and the query go in together and the solver answers, with trace available when the result is surprising enough to need the derivation.
02
Building up a domain you will keep asking questions about
Stable rules are saved once under a name and pulled into later runs, so a scheduling or game domain stops being retyped into every prompt.
03
Retiring rules that no longer describe the world
The stored set can be reviewed by name and description, and anything that has gone stale is removed so later queries do not quietly rely on it.
5 tools
Counted once with only startup setup. Other versions, modes or optional credentials may expose a different list.
execute_prolog
Execute Prolog code and return reasoning results.
Write Prolog facts and rules, then run a query against them. Supports CLP(FD) constraints, negation-as-failure, and all standard SWI-Prolog features.
get_rule_base
Retrieve the Prolog source of a saved rule base.
list_rule_bases
List all saved rule bases with description and tags.
Returns `{"rule_bases": [{"name": str, "description": str, "tags": list[str]}, ...]} sorted by name. Metadata is extracted from the leading % description: / % tags:` comments of each rule base file (see §4.10).
save_rule_base
Save a named rule base containing Prolog rules that can be reused across `execute_prolog` calls.
Use this for stable, reusable knowledge (e.g. `piece_moves for chess piece movement rules). For one-time facts, include them directly in prolog_code` instead.
BEFORE YOU INSTALL
What to know first
Tools you get
5 tools when we connected.
Requires
- uvx on PATH
- Python ≥ 3.10
- SWI-Prolog 9.0 or newer, installed and on PATH
Where it runs
Both documented clients run the server on your machine.
CONTEXT COST
Light Standing cost
Rated by the size of the 5 tool definitions that ride along on every turn, not by how many there are
INSTALLATION
Connect your client
Server setup
TERMINAL
pip install prolog-reasonerMCP Clients
Claude Desktop, Claude Code
JSON
{
"mcpServers": {
"prolog-reasoner": {
"command": "uvx",
"args": [
"prolog-reasoner"
]
}
}
}
