CodeLogician applies neurosymbolic AI to translate source code into precise mathematical logic, striving to create a formal model of the program's behavior that's functionally equivalent to the original source code.
Project description
CodeLogician
CodeLogician is the neurosymbolic agentic governance framework for AI-powered coding.
It helps your coding agent think logically about the code it's producing and test cases it's generating. The fundamental flaw that all LLM-powered assistants have is the reasoning they're capable of is based on statistics, while you need rigorous logic-based automated reasoning.
- Generated code is based on explainable logic, not pure statistics
- Generated test cases are generated come with quantitative coverage metrics
- Generated code is consistent with the best security practices leveraging formal verification
To run CodeLogician, please obtain an Imandra Universe API key available (there's a free starting plan)
at Imandra Universe and make sure it's available in your environment as IMANDRA_UNI_KEY.
Three typical workflows:
- DIY mode - this is where your agent (e.g. Grok) uses the CLI to:
- Learn how to use IML/ImandraX via
doccommand (e.g.codelogician doc --help) - Synthesizes IML code and uses the
evalcommand to evaluate it - If there're errors, use
codelogician doc view errorscommand to study how to correct the errors and re-evalute the code
- Agent/multi-agent mode - CodeLogician IML Agent is a Langgraph-based agent for automatically formalizing source code into Imandra Modeling Language (IML).
- With
agentcommand you can formalize a single source code file (e.g.codelogician agent PATH_TO_FILE) - With
multiagentcommand you can formalize a whole directory (e.g.codelogician agent PATH_TO_DIR)
- Server
- This is a "live" and interactive version of the
multiagentcommand - It monitors the filesystem and fire's off formalization tasks on source code updates as necessary
- You can start the server and connect to it with the TUI (we recommend separate terminal screens)
Learn more at CodeLogician!
To get started,
codelogician --help
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file codelogician-2.0.0b10.tar.gz.
File metadata
- Download URL: codelogician-2.0.0b10.tar.gz
- Upload date:
- Size: 614.4 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: uv/0.8.22
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
afc3dd7168377715bdfccbcd12767bd62eb163ba2c9588064c5fd3491fb3591e
|
|
| MD5 |
476d5d2d5b09bc71fc30d94ac39b2b42
|
|
| BLAKE2b-256 |
ae67b7d58331dc47185d39bcf57a8b7bdf7363cb0a365a810f18e796246a3f14
|
File details
Details for the file codelogician-2.0.0b10-py3-none-any.whl.
File metadata
- Download URL: codelogician-2.0.0b10-py3-none-any.whl
- Upload date:
- Size: 730.1 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: uv/0.8.22
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
f738b53e21dc23edcb68c81ad85775136f71d6d947fb204fbc9e2ac63d96fa58
|
|
| MD5 |
459dfad9dba425c92576bad638d7cf6b
|
|
| BLAKE2b-256 |
1685c6ef9f3be0cfd56b264aa1808d176a6c2258a2c35c7e38bff21f532c6d54
|