PRAXIST
Autonomous research system for measurable, computer-executable research.
PRAXIST is an autonomous research system that transforms measurable, runnable projects into persistent, multi-generation experimentation loops. It coordinates parallel research agents that explore competing hypotheses, evaluate results against task-defined metrics, and synthesize evidence across generations to guide optimization. Designed to work through Codex and Claude Code, PRAXIST adds research orchestration, evidence protocols, and lifecycle control while the interactive agent handles project understanding and tool usage. The system maintains parallel research peers, durable evidence lanes, and multi-metric evaluation to systematically search solution spaces when the optimal path forward is uncertain.
Key Features
Use Cases
- 01Optimizing machine learning model architectures and hyperparameters when the baseline runs but the improvement path is unclear
- 02Exploring algorithm design spaces with measurable performance metrics across competing implementation strategies
- 03Running continuous experimentation loops on engineering simulations with multi-objective optimization constraints
- 04Conducting systematic research on runnable codebases where manual iteration is time-intensive and the search space is large
- 05Generating auditable evidence packages for research decisions, including negative results that rule out unproductive directions
- 06Automating multi-generation hypothesis testing with persistent evidence synthesis and provenance tracking
Related Plugins
View moreandrej-karpathy-skills
A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.
claude-mem
Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More
Understand-Anything
Graphs that teach > graphs that impress. Turn any code into an interactive knowledge graph you can explore, search, and ask questions about. Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more.
rtk
CLI proxy that reduces LLM token consumption by 60-90% on common dev commands. Single Rust binary, zero dependencies
PRAXIST — FAQ
What is PRAXIST and what does it do?+
PRAXIST is an autonomous research system that turns already-runnable projects with measurable objectives into continuous, multi-generation experimentation loops. Parallel research agents develop candidate solutions, evaluators convert results into structured evidence, and a planning panel synthesizes findings to guide the next generation until convergence or budget exhaustion.
How do I install PRAXIST for Claude Code?+
Use the host-specific one-line installation command provided in the documentation's installation guide. Alternatively, open Codex and ask it to install and configure PRAXIST using the packaged out-of-box-experience runbook, stopping after readiness checks pass.
Which AI clients work with PRAXIST?+
PRAXIST is designed to work through Codex and Claude Code as the interactive interface. Direct CLI operation is also available without an agent client for manual control, monitoring, and inspection.
Do I need API keys to use PRAXIST?+
No API key is required in Codex-native mode, which uses your authenticated Codex session. For sustained research, you can optionally provide supported model provider API keys through the masked credential setup. API costs depend on the provider, model selection, parallelism level, and generation count.
Is PRAXIST free to use?+
PRAXIST is licensed under the Fair Source License 1.0. Organizations with aggregate annual revenue below US$1 million may use it commercially at no charge. Organizations exceeding that threshold must contact Sapient Intelligence Pte Ltd for a commercial license. The revenue threshold does not apply to qualifying academic research at educational and nonprofit research institutions.
What are the prerequisites for running PRAXIST research?+
You need CPython 3.11+, a runnable project with executable code, measurable evaluation metrics, and a defined baseline. PRAXIST will not download datasets, invent simulators, or fabricate baselines—all project assets must already be in place and working before starting research.
How do I install PRAXIST?+
Open the source repository on GitHub and follow its README. PRAXIST is a plugin — MCP Agents Market links you directly to the official repo.
Is PRAXIST free?+
PRAXIST is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.