Metadata-Version: 2.4
Name: praxist
Version: 0.5.0
Summary: Praxist — multi-agent, multi-generation autonomous ML research system
Project-URL: Documentation, https://praxist.sapient.inc/en/docs
License: # Fair Source License Agreement (Version 1.0)
        
        This License Agreement (the "Agreement") is a legal agreement between you (an
        individual or legal entity, the "Licensee") and Sapient Intelligence Pte Ltd
        (the "Licensor"), governing Praxist by Sapient Intelligence accompanied
        herewith (including source code, object code, documentation and related
        materials, collectively the "Software"). By installing, copying or using the
        Software, you agree to be bound by the terms of this Agreement. If you do not
        agree, you may not install or use the Software.
        
        ## 1.1 Definitions
        
        ### 1.1.1 "Software"
        
        means Praxist by Sapient Intelligence developed by the Licensor, including all
        source code, object code, configuration files, pre-trained weights (if any),
        documentation, examples and related materials, in whatever form or medium they
        are provided.
        
        ### 1.1.2 "Licensee" / "You"
        
        means the individual or legal entity accepting the terms of this Agreement. If
        you accept this Agreement on behalf of a legal entity, you represent and
        warrant that you have full authority to bind such entity to this Agreement.
        
        ### 1.1.3 "Annual Revenue"
        
        means the worldwide aggregate gross revenue (not net profit or revenue after
        deduction of costs) of the Licensee and its Affiliates (as defined in Section
        1.1.4) arising from any commercial activities during the most recently
        completed fiscal year. If the Licensee has not operated for a full fiscal year,
        its annual revenue shall be annualized by multiplying its average monthly
        revenue since its establishment by twelve (12). All amounts shall be
        denominated in United States Dollars (USD); non-USD revenue shall be converted
        at the closing mid-rate published by Reuters on the last day of the relevant
        fiscal year. If Reuters does not publish a rate for the relevant currency on
        such date, the closing mid-rate published by Reuters on the most recent trading
        day for which such rate is available shall apply. Should Reuters permanently
        cease to publish such exchange rate data, the parties shall negotiate to select
        a comparable closing mid-rate published by another internationally recognized
        third-party financial information service provider as a substitute benchmark.
        
        ### 1.1.4 "Affiliate"
        
        means any legal entity that, directly or indirectly, controls the Licensee, is
        controlled by the Licensee, or is under common control with the Licensee.
        "Control" means the direct or indirect ownership of more than fifty percent
        (50%) of the voting equity or interests of such entity, or the right to direct
        the management and policies of such entity through contract or otherwise.
        
        ### 1.1.5 "Generate" / "Generated Output"
        
        means any model parameter updates, model weight files, inference outputs or
        optimization results output or produced by the Software upon use by the
        Licensee.
        
        ### 1.1.6 "Commercial License"
        
        means a paid license agreement to be separately executed between the Licensee
        and the Licensor by reason of the Licensee's Annual Revenue reaching or
        exceeding the Revenue Threshold (as defined in Section 1.3.1).
        
        ### 1.1.7 "Effective Date"
        
        means the date on which the Licensee first installs, copies or uses the
        Software.
        
        ## 1.2 Grant of License
        
        ### 1.2.1 Free License
        
        Subject to all terms and conditions of this Agreement (including, without
        limitation, the Revenue Threshold condition under Section 1.3), provided that
        the Licensee satisfies the "Revenue Threshold Condition" (as defined in
        Section 1.3.1), the Licensor
        hereby grants to the Licensee a worldwide, non-exclusive, non-sublicensable
        (except for the purpose set out in Section 1.2.3), non-transferable (except as
        set out in Section 1.2.4) license to:
        
        (a) use, copy and modify the Software for internal business purposes;
        
        (b) create derivative works based on the Software for internal business
        purposes; and
        
        (c) deploy and run the Software within its own organization, whether by way of
        online service or on-premises installation.
        
        ### 1.2.2 Source Code Availability
        
        The Licensor shall make the complete source code of the Software publicly
        available on a designated code hosting platform. The Licensee is entitled to
        view, download and modify the Software based on such source code, provided that
        the modified version (the "Modified Version") remains subject to all terms and
        conditions of this Agreement.
        
        ### 1.2.3 Distribution to Third Parties
        
        The Licensee may not distribute, sublicense, sell, rent, lease or transfer the
        Software or any Modified Version thereof to any third party as a standalone
        product or service, unless prior written consent of the Licensor has been
        obtained. This restriction does not prohibit the Licensee from providing the
        functionality of the Software to end users as part of its own services,
        provided that such services are not independently sold under the name of
        "Praxist by Sapient Intelligence."
        
        ### 1.2.4 Transfer Restrictions
        
        The Licensee may not assign this Agreement or any rights or obligations
        hereunder without the prior written consent of the Licensor. The Licensor shall
        not unreasonably withhold such consent. Any assignment in violation of this
        Section shall be void.
        
        ### 1.2.5 Output Attribution Obligation
        
        The Licensee is not required to comply with this attribution obligation in
        respect of output generated by the product for its own internal use. If the
        Licensee publicly releases, distributes or otherwise makes available to any
        third party any output generated by the product, the Licensee must retain the
        Licensor's product name attribution in a reasonably prominent position in such
        output, and may not remove, obscure or alter such attribution. Such attribution
        may be placed in the footer, explanatory notes, metadata or accompanying
        documentation of the output; the Licensor may provide recommended attribution
        wording in writing. The foregoing attribution obligation shall continue to
        apply even if the Licensee modifies, edits or otherwise reprocesses the output.
        
        ## 1.3 Revenue Threshold and Commercial License
        
        ### 1.3.1 Revenue Threshold
        
        "Revenue Threshold" means that the aggregate Annual Revenue (as defined in
        Section 1.1.3) of the Licensee and its Affiliates reaches or exceeds One
        Million United States Dollars (USD $1,000,000.00), denominated in USD.
        
        ### 1.3.2 Condition of the Free License
        
        The free license under Section 1.2.1 shall be valid only if the aggregate
        Annual Revenue of the Licensee and its Affiliates is below the Revenue
        Threshold. Once the Licensee's Annual Revenue reaches or exceeds the Revenue
        Threshold, the free license shall automatically lapse, and the Licensee shall
        contact the Licensor within thirty (30) days to negotiate a Commercial License.
        
        ### 1.3.3 Obligation to Negotiate a Commercial License
        
        Once the Licensee's Annual Revenue reaches or exceeds the Revenue Threshold:
        
        (a) the Licensee shall notify the Licensor in writing within thirty (30)
        calendar days through the Licensor's designated contact channels (see the end
        of this Agreement);
        
        (b) the parties shall negotiate the terms of a Commercial License in good
        faith within sixty (60) calendar days after such notice; and
        
        (c) if the parties fail to reach a Commercial License agreement within ninety
        (90) calendar days, this free license shall automatically terminate, and the
        Licensee shall cease using the Software and delete all copies of the Software
        from all of its systems.
        
        ### 1.3.4 Revenue Reporting Obligation
        
        The Licensor is entitled, at intervals of every twelve (12) months, to require
        the Licensee to provide a revenue statement signed by an authorized
        representative, confirming whether the Licensee's Annual Revenue remains below
        the Revenue Threshold. The Licensee shall provide such statement within thirty
        (30) calendar days of receiving the request. If the Licensee refuses or fails
        to provide such statement, the Licensor is entitled to terminate this license.
        
        ### 1.3.5 Good Faith Presumption
        
        The Licensor and the Licensee shall each deal with any dispute arising from the
        Revenue Threshold in good faith. If the Licensee has a reasonable dispute as to
        the basis of revenue calculation, the parties shall maintain the status quo
        during the pendency of such dispute and neither party shall unilaterally
        terminate the license.
        
        ### 1.3.6 Academic Institution Exemption
        
        The Revenue Threshold restrictions under this Agreement shall not apply to
        institutions of higher education, public research institutes and other
        non-profit academic research institutions. This exemption applies solely to
        such institution's own use for teaching, academic research, scientific
        experimentation and non-profit publication of academic results. If any such
        institution uses the Software through its controlled for-profit subsidiary,
        spin-off enterprise or technology transfer platform, or for fee-based
        commercial services offered to the market, such entity shall not be entitled
        to this exemption and shall fully comply with all Revenue Threshold-related
        provisions of this Agreement.
        
        ## 1.4 Data Collection
        
        Product-usage data collection is optional and is not a condition of the license
        granted under Section 1.2. Accepting this Agreement, installing the Software or
        using the Software does not constitute consent to product-usage collection.
        
        The Licensor may collect the bounded product-usage data described in the User
        Agreement and its Praxist User Data Collection Notice only after the Licensee
        provides a separate, explicit opt-in through a supported consent flow. If
        consent is unset, denied or withdrawn, the Software shall not collect or send
        product-usage events. Declining or withdrawing consent shall not reduce the
        Software's research functionality. The Licensee may withdraw consent at any
        time through the mechanism documented in the User Data Collection Notice.
        
        The specific data fields, processing rules, retention period and withdrawal
        behavior are governed by the User Agreement and its User Data Collection
        Notice. If the Licensee integrates the product to provide services to
        third-party end users, the Licensee shall ensure compliance with applicable
        data protection laws and regulations at its own responsibility.
        
        ## 1.5 Restrictions
        
        Except as expressly permitted under this Agreement or as mandatorily permitted
        by applicable law, the Licensee may not:
        
        - remove, modify or obscure any copyright notices, trademarks, license terms or
          attribution information in the Software or any copy thereof;
        - use the Software for any purpose that violates applicable laws and
          regulations, including but not limited to generating illegal content,
          infringing the intellectual property rights of others, or conducting
          discriminatory or harmful AI training;
        - use the Software or its name/trademark in a manner that misleads others into
          believing that the Licensee is the developer of the Software;
        - reverse engineer, decompile or disassemble the Software, except to the extent
          expressly permitted by applicable law and not capable of being excluded by
          contract; or
        - use the Software for military, surveillance or large-scale human rights
          violations.
        
        ## 1.6 Intellectual Property
        
        The Software and all intellectual property rights therein worldwide (including
        but not limited to copyrights, patent rights, trademark rights, trade secrets
        and know-how) are the exclusive property of the Licensor. This Agreement does
        not transfer to the Licensee any ownership of intellectual property rights,
        and grants only a limited license of use. In respect of Modified Versions
        created by the Licensee based on the Software, the Licensee owns the copyright
        in the portions independently created by it, provided that such Modified
        Version as a whole remains subject to this Agreement, and the Licensee hereby
        grants to the Licensor a perpetual, irrevocable, royalty-free, worldwide
        license to use any suggestions for improvement or patches submitted by the
        Licensee to the Licensor.
        
        ## 1.7 Disclaimer of Warranties
        
        The Software is provided "AS IS", without warranties of any kind, express or
        implied, including but not limited to the implied warranties of
        merchantability, fitness for a particular purpose, title and non-infringement.
        The Licensor does not warrant that the Software is error-free, that operation
        will be uninterrupted, or that all defects will be corrected. The Licensee
        assumes the entire risk of using the Software. The Software is an AI model
        tuning tool, and its output may contain bias, errors or inaccuracies. The
        Licensee should independently verify results prior to deploying any model
        tuned through the Software.
        
        ## 1.8 Limitation of Liability
        
        To the maximum extent permitted by applicable law, in no event shall the
        Licensor be liable for any indirect, incidental, special, punitive or
        consequential damages (including but not limited to loss of profits, loss of
        data, business interruption or loss of goodwill) arising out of or in
        connection with this Agreement or the use of or inability to use the Software,
        whether based on any theory of liability (contract, tort including negligence,
        strict liability or otherwise), even if advised of the possibility of such
        damages. The aggregate liability of the Licensor arising out of this Agreement
        shall in no event exceed the amount of license fees actually paid by the
        Licensee to the Licensor for the Software during the twelve (12) months
        preceding the event giving rise to the claim. If the Licensee's use was free of
        charge during such period, the aggregate liability of the Licensor shall not
        exceed One Hundred United States Dollars (USD $100.00).
        
        ## 1.9 Termination
        
        ### 1.9.1 Termination by the Licensee
        
        The Licensee may terminate this Agreement at any time by ceasing use of the
        Software and deleting all copies thereof from its systems.
        
        ### 1.9.2 Termination by the Licensor
        
        If the Licensee materially breaches any provision of this Agreement and fails
        to cure such breach within thirty (30) calendar days of receiving written
        notice of the breach from the Licensor, the Licensor may terminate this
        Agreement. The Licensee's failure to notify the Licensor within the period
        prescribed in Section 1.3.3(a) after its Annual Revenue reaches or exceeds the
        Revenue Threshold constitutes an incurable material breach, and the Licensor
        may terminate this Agreement immediately upon written notice.
        
        ### 1.9.3 Effect of Termination
        
        Upon termination of this Agreement: (a) all licenses granted under this
        Agreement shall terminate immediately; (b) the Licensee shall cease using the
        Software and permanently delete all copies of the Software and any Modified
        Versions thereof from all systems under its control; and (c) Section 1.6
        (Intellectual Property), Section 1.7 (Disclaimer of Warranties), Section 1.8
        (Limitation of Liability) and Section 1.10 (Governing Law) shall survive such
        termination.
        
        ## 1.10 Governing Law and Dispute Resolution
        
        ### 1.10.1 Governing Law
        
        This Agreement shall be governed by and construed in accordance with the laws
        of Singapore, without regard to its conflict of laws principles. The United
        Nations Convention on Contracts for the International Sale of Goods shall not
        apply to this Agreement.
        
        ### 1.10.2 Dispute Resolution
        
        Any dispute arising out of or in connection with this Agreement shall first be
        resolved through friendly negotiation between the parties. If such negotiation
        fails, either party may submit the dispute to the Singapore International
        Arbitration Centre for resolution.
        
        ### 1.10.3 Severability
        
        If any provision of this Agreement is held by a court of competent jurisdiction
        to be invalid or unenforceable, such provision shall be enforced to the maximum
        extent permitted by law, and the remaining provisions shall continue in full
        force and effect.
        
        ### 1.10.4 Entire Agreement
        
        This Agreement constitutes the entire agreement between the parties with
        respect to the subject matter hereof. Checking the confirmation box for this
        Agreement shall be deemed acknowledgment and execution thereof.
        
        - **Email:** praxist@sapient.inc
        - **Publication Date:** 08/28/2026
License-File: LICENSE.md
Requires-Python: >=3.11
Requires-Dist: jinja2>=3.1
Requires-Dist: pydantic<3,>=2.7
Requires-Dist: pyyaml>=6.0
Provides-Extra: agents
Requires-Dist: anthropic>=0.34; extra == 'agents'
Requires-Dist: claude-agent-sdk==0.2.136; extra == 'agents'
Requires-Dist: httpx>=0.27; extra == 'agents'
Requires-Dist: mcp>=1.0; extra == 'agents'
Requires-Dist: openai>=1.40; extra == 'agents'
Requires-Dist: pdf2image>=1.17; extra == 'agents'
Requires-Dist: pypdf>=4.0; extra == 'agents'
Requires-Dist: pytesseract>=0.3; extra == 'agents'
Requires-Dist: requests>=2.32; extra == 'agents'
Requires-Dist: trafilatura>=1.12; extra == 'agents'
Provides-Extra: codex
Requires-Dist: claude-agent-sdk==0.2.136; extra == 'codex'
Requires-Dist: codex-relay==0.5.5; extra == 'codex'
Requires-Dist: mcp>=1.0; extra == 'codex'
Requires-Dist: openai-codex==0.147.0; extra == 'codex'
Provides-Extra: docs
Requires-Dist: mkdocs-material<10,>=9.7; extra == 'docs'
Requires-Dist: mkdocstrings[python]<2,>=1.0; extra == 'docs'
Provides-Extra: product-usage-server
Requires-Dist: alembic<2,>=1.16; extra == 'product-usage-server'
Requires-Dist: fastapi<1,>=0.116; extra == 'product-usage-server'
Requires-Dist: psycopg[binary]<4,>=3.2; extra == 'product-usage-server'
Requires-Dist: sqlalchemy<3,>=2.0; extra == 'product-usage-server'
Requires-Dist: uvicorn<1,>=0.35; extra == 'product-usage-server'
Provides-Extra: storage
Requires-Dist: boto3>=1.34; extra == 'storage'
Requires-Dist: botocore>=1.34; extra == 'storage'
Description-Content-Type: text/markdown

<p align="center">
  <img src="docs/assets/brand/praxist-banner.svg" alt="Praxist" width="800">
</p>

<h1 align="left">Praxist: meet your personal R&amp;D team</h1>

<p align="center">
  <a href="../../actions/workflows/ci.yml"><img src="../../actions/workflows/ci.yml/badge.svg" alt="CI"></a>
  <a href="https://praxist.sapient.inc/en/docs"><img src="https://img.shields.io/badge/docs-open-334155" alt="Documentation"></a>
  <a href="https://arxiv.org/abs/2608.25955"><img src="https://img.shields.io/badge/arXiv-2608.25955-B31B1B.svg" alt="arXiv paper"></a>
  <a href="https://discord.gg/sapient"><img src="https://img.shields.io/badge/Discord-Join-5865F2?logo=discord&amp;logoColor=white" alt="Discord"></a>
  <img src="https://img.shields.io/badge/Python-3.11%2B-3776AB" alt="Python 3.11+">
</p>

Praxist is an autonomous research system for measurable, computer-executable
research. It coordinates parallel research peers, task-owned evaluation,
durable evidence, and generation-to-generation synthesis.

Praxist treats research as a persistent process rather than a sequence of
disconnected prompts. Use it when a project already runs and its objective is
measurable, but the best path forward is still unknown.

<p align="center">
  <img src="docs/assets/figures/praxist-overview.svg" alt="Praxist architecture" width="920">
</p>

## Install Praxist

Install the complete runtime integrations and finish first-use setup with one
command:

```bash
python3 -m pip install --index-url https://pypi.org/simple "praxist[agents,codex]" && praxist setup --interactive --install-skills codex
```

The local wizard covers the Fair Source License, User Agreement, privacy,
runtime profile, masked credentials, Codex skills, writable examples,
and readiness checks. It does not select a research project or launch a run.
For Claude Code, use the
[host-specific one-line command](docs/getting-started/installation.md#install-and-configure).

For an agent-managed installation, open Codex:

```bash
codex --yolo
```

Then ask it to install and configure Praxist using the packaged OOBE runbook,
and to stop after readiness checks.

Before starting research, read the [Quickstart](docs/getting-started/quickstart.md)
and [Your First Task](docs/getting-started/first-task.md). They describe the
separate takeover step and the project contract it creates.

Choose **Codex-native mode** to use an existing Codex subscription without an
API key. For sustained research, Praxist generally favors
[open-source model APIs](docs/guides/open-source-model-apis.md) with a high
observed cache-hit rate. The setup wizard also supports other API-backed
profiles.

## Use Praxist Through Codex

We recommend Codex as the interface for operating Praxist. Praxist is not a
replacement for Codex: Codex remains the interactive agent that understands
your project, communicates with you, and uses development tools. Praxist adds
the persistent research loop, parallel peers, evidence protocols, scheduling,
and lifecycle control.

After installation, open Codex in the root of an already runnable research
project and invoke `$praxist-takeover`. The takeover skill inspects readiness,
creates or repairs the task harness, validates its evaluator and evidence
contract, and launches the run after the required gates pass. A precise brief
produces a better research plan; include the objective, metrics, constraints,
resources, exploration choices, and whether launch is authorized.

<details>
<summary>Example takeover brief</summary>

```text
$praxist-takeover

Treat the current directory as the existing runnable research project. Verify
the baseline and its evaluation path before changing anything.

Optimize <primary metric and direction> while preserving <key constraints>.
Use <peer count> peers for up to <generation count> generations within
<time or cost budget>. Use the runtime and model provider selected during
setup. <Allow or disable> literature search, <enable or disable> QD, and
<enable or disable> generation-zero DIG.

Do not download new datasets or replace required project assets. Build a
separate task harness with explicit metric directions, baseline provenance,
protocol-integrity checks, evidence maturity rules, and justified retention
lanes. After readiness checks pass, <launch immediately in detached mode / ask
for confirmation>. Report the task path, run ID, evidence contract, generation
close policy, and monitor command.
```

</details>

Other bundled skills:

| Skill | Purpose |
|---|---|
| `praxist-takeover-codex` | No-key takeover using the saved Codex login |
| `praxist-onboarding` | Explain Praxist and inspect local readiness |
| `praxist-task-initialization` | Build or repair a task harness without launching |
| `praxist-interactive-task-init` | Design a task through confirmation-first setup |
| `praxist-control` | Start, stop, resume, monitor, and inspect runs |
| `praxist-diagnostic` | Diagnose run health and produce reports |
| `praxist-scientific-research` | Gather sourced literature and benchmark context |
| `praxist-runtime-install` | Install or repair runtime dependencies and credentials |
| `terminal-line-plot` | Draw metric trends in the terminal |

See [Agent Skills](docs/user-guide/skills.md) for invocation syntax and the
generated [Skills Reference](docs/reference/skills.md) for the complete
contracts.

## What Praxist Provides

| Capability | Purpose |
|---|---|
| Parallel research peers | Explore competing hypotheses and implementations concurrently |
| Multi-generation synthesis | Carry useful evidence and strategy into later generations |
| Durable evidence lanes | Preserve candidates through incubator, frontier, and Gems state |
| Multi-metric evaluation | Rank task-defined evidence, including Pareto-optimal tradeoffs |
| [Quality-Diversity (QD)](docs/guides/qdig-cohort-allocator.md) and optional [Deep Innovation Gate (DIG)](docs/guides/deep-innovation-gate.md) | Maintain diversity without forcing one exploration policy |
| Central resource scheduling | Adapt experiment admission to observed resource pressure |
| Resume, replay, and monitoring | Keep long-running research inspectable and recoverable |
| Plugin boundaries | Support multiple runtimes, providers, tools, budgets, and workflows |

## Praxist And The Task Project

| Praxist owns | The task project owns |
|---|---|
| Research orchestration, lifecycle, evidence protocols, replay, scheduling, and extension interfaces | Research objective, executable code, evaluator, metrics, baselines, prompts, roles, and domain constraints |

Praxist contains no task-specific scientific assumptions. A task remains the
single source of truth for what should be tested and what counts as valid evidence.

## Operate A Run

```bash
praxist status --json
praxist --monitor --latest
praxist stop <run_id>
praxist resume <run_dir>
```

`Ctrl-C` closes only the monitor; it does not stop the research run.

## Examples And Templates

```bash
praxist examples list
praxist examples install rocket_booster_recovery
praxist examples install rocket_booster_recovery_rust
```

Complete examples are writable reference projects. `templates/tasks/` contains
replaceable scaffolding for building new task harnesses. The two Rocket Booster
Recovery examples demonstrate the same research problem through Python/JAX and
native Rust implementations.

## Requirements

| Status | Requirement |
|---|---|
| Required | CPython 3.11+ |
| Required to launch research | A runnable project with measurable evaluation |
| Required for skill-driven operation | Codex or Claude Code; direct CLI operation remains available without either |
| Authentication: choose one | A saved Codex login for Codex-native mode, or a supported provider API key |
| Continuously release-tested | Linux on CPython 3.11 and 3.12 |
| Compatibility target | macOS and other CPython 3.11+ environments; run `praxist doctor` before research |

Task-specific datasets and compute dependencies remain owned by the task project.

See the [platform support matrix](docs/operations/platform-support.md) for the
difference between release-qualified hosts and compatibility targets.

## Documentation

Read the **[Praxist documentation](https://praxist.sapient.inc/en/docs)**
or open it with:

```bash
praxist docs
```

No local documentation server is required.

## Contributing To Praxist

Read the [contribution guide](.github/CONTRIBUTING.md) together with the
[Code of Conduct](.github/CODE_OF_CONDUCT.md) before participating. Source
maintainers should also follow the architecture and maintenance contract in
[AGENTS.md](AGENTS.md). See the [Privacy Notice](docs/legal/PRIVACY.md) for
Praxist's data-handling terms.

Contact: praxist@sapient.inc

## FAQ

<details>
<summary>Show questions and answers</summary>

### Q1. What is Praxist?

Praxist is an autonomous research system for measurable research problems that
can be executed on a computer. It turns an already runnable project into a
continuous, evidence-driven research run.

Across successive generations, parallel research agents develop candidate
solutions; evaluators convert results into structured evidence; and a planning
panel synthesizes that evidence into the research agenda for the next
generation. The cycle continues until the search converges or the budget is
exhausted.

You provide a runnable project and a measurable objective. Praxist orchestrates
the research process that searches for the best-performing solution.

### Q2. How is Praxist different from manual tuning or AutoML?

AutoML tunes parameters within a predefined search space. Praxist runs the full
research loop.

Parallel research agents can change methods, architectures, and strategies.
Evidence from evaluation shapes the agenda for the next generation, while the
[Deep Innovation Gate (DIG)](docs/guides/deep-innovation-gate.md) and
[Quality-Diversity (QD)](docs/guides/qdig-cohort-allocator.md) allocation help
the system escape local optima.

Praxist is closer to a self-directing research team than a search tool. If your
researchers are already iterating on a problem manually, Praxist takes over the
iteration loop itself.

### Q3. Is my project a good fit for Praxist?

Praxist delivers the most value when three conditions are met:

- **The objective is measurable:** there is at least one metric that
  meaningfully distinguishes better from worse, with a clear optimization
  direction.
- **The project already runs:** the baseline code, environment, and required
  data or simulator are in place and work without Praxist.
- **The best path forward is unknown.**

If a prerequisite is missing, Praxist stops and tells you exactly what is
needed. It will not silently download unspecified datasets, invent a simulator,
or fabricate baseline performance. That is a deliberate design principle.

### Q4. Do I need an API key, and what will it cost?

No API key is required in Codex-native mode; Praxist uses your authenticated
Codex session. We also recommend using your own API key to access supported
model APIs.

API costs are set by the provider and vary by model and usage. Total cost also
depends on parallelism, the number of generations, and evaluation runtime. For
cost-sensitive runs, start with a small representative workload before scaling
up.

### Q5. How does Praxist protect my code and data?

Praxist provides three layers of protection:

- **Project isolation:** Praxist does not modify your original project. Run
  artifacts are stored separately.
- **Credentials:** API keys are entered through a masked local prompt and are
  not exposed in commands, shell history, or conversations.
- **Data collection:** Praxist does not collect data used in your experiments.
  It collects only limited system-level operational information, which you can
  disable at any time.

### Q6. How can I trust that a reported improvement is real?

Praxist uses three safeguards:

- **Preregistration:** Metrics, evaluation protocols, baselines, and acceptance
  thresholds are defined before the run.
- **Consistent evaluation:** Every candidate is measured through the same
  evaluator, and invalid or suspicious results are excluded.
- **End-to-end provenance:** Every reported improvement includes the evidence
  and lineage needed to inspect and reproduce it.

We recommend reviewing what the selected solution changed and testing it again
in your own environment. Praxist's results are designed to be verifiable, and
your own validation should be the final test.

### Q7. What if Praxist does not improve the result?

Praxist does not guarantee a specific metric improvement. It provides a
rigorous research process and auditable evidence.

If a run does not meet its target, you still receive a negative-result evidence
package, an audit report, and recommendations on whether to stop or redirect
the research.

A negative result can still be valuable: it rules out tested approaches with
evidence and helps prevent further investment in an unproductive direction.

### Q8. Is Praxist open source, and what terms apply to its outputs?

Praxist is licensed under the Fair Source License Agreement 1.0. The precise
description is source-available: the complete source code is publicly
available and may be viewed, downloaded, and modified. Subject to the license
terms, Praxist may be used for internal business purposes and deployed within
your own organization.

Organizations with aggregate annual revenue, including revenue from
affiliates, below US$1 million may use Praxist commercially at no charge. Once
annual revenue reaches or exceeds that threshold, the organization must
contact the Licensor, Sapient Intelligence Pte Ltd, to negotiate a Commercial
License.

The revenue threshold does not apply to qualifying teaching and academic
research conducted by institutions of higher education, public research
institutions, and nonprofit academic research organizations.

**Generated outputs:** no attribution is required for internal use. If an
output is published externally or otherwise made available to third parties,
the product-name attribution "Praxist by Sapient Intelligence" must be
retained.

This FAQ is a summary only. If it conflicts with the Fair Source License
Agreement 1.0, the terms of the license agreement control.

</details>

## Citation

If you use Praxist in your research, please cite:

```bibtex
@misc{li2026praxistexperimentalartifactssolution,
      title={Praxist: From Experimental Artifacts to Solution Lineages},
      author={Jin Li and Ahmed Murtadha and Zhiyu Wang and Qiwen Chen and William Chen and Yifei Wu and Guan Wang and Andy L. Siy and Jiayi Yang and Mengsha Huang and Wenhao Li and Yixuan Liu and Shuailin Pan and Mingli Yuan and Sen Song and Yuhao Sun},
      year={2026},
      eprint={2608.25955},
      archivePrefix={arXiv},
      primaryClass={cs.MA},
      url={https://arxiv.org/abs/2608.25955},
}
```
