scitex-dev

What it is and what it does

scitex-dev is a CLI and Python library that centralizes developer operations for a large ecosystem of related packages. It solves the coordination problem of keeping ~70 packages in sync: enforcing naming conventions, API rules, and project structure through an audit system with ~130 numbered rules; aggregating skills (workflow guides) from all packages for AI agent discovery; detecting and fixing version mismatches across the ecosystem; and performing bulk renames with cross-reference updates and safety checks. The package ships as a single install with no feature menu—everything (CLI, MCP server, icon rendering, skills) is included.

It provides four interfaces: a CLI (primary), a Python API for version management and CI/CD checks, an MCP server for integration with Claude, and a skills system that exports workflow guides to ~/.claude/skills/scitex/. The linter engine aggregates per-package rule plugins, allowing rules to ship alongside the code they enforce. Configuration lives in optional local-state directories (~/.scitex/dev/ and <proj-root>/.scitex/dev/) and can be overridden by environment variables or CLI flags.

Use it for:

• Run ecosystem-wide audits to catch version drift, README inconsistencies, and API violations across 70+ packages in one command.
• Aggregate and export skills from all SciTeX packages so AI agents can discover and use them without manual curation.
• Detect and fix version mismatches across the ecosystem, then sync changes to all hosts in a single operation.
• Perform atomic bulk renames across multiple repos with automatic cross-reference updates and dry-run preview.
• Enforce linting rules tied to specific APIs, ensuring that renamed functions and their corresponding lint rules stay in sync.
• Query documentation and search across all ecosystem packages from the CLI or Python API.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Provides unified developer tooling for the SciTeX ecosystem: audits ~70 packages against ~130 rules, aggregates skills for AI agents, manages coordinated releases and bulk renames, and runs a linter that enforces API consistency across the ecosystem.

Yes, if you maintain or develop within the SciTeX ecosystem—it directly solves the coordination and consistency problems described in its problem table. No, if you are outside that ecosystem: the audit rules, skills system, and ecosystem commands are tightly coupled to SciTeX conventions and will not transfer. The AGPL-3.0 license also requires your project to be open-source under compatible terms or kept private.

Install

scitex-dev on PyPI

pip install scitex-dev

uv add scitex-dev

poetry add scitex-dev

Installing scitex-dev

Before you install

Low install friction: pure Python wheel with 15 runtime dependencies including click, pyyaml, textual, and rich. Active maintenance with a release 1 day old. Requires Python 3.9+.

License in practice

AGPL-3.0-only: using this package in your own project requires you to license that project under AGPL-3.0 or compatible terms, or keep it private. Suitable for internal tooling and open-source projects already under copyleft licenses.

Quickstart

pip install scitex-dev

scitex-dev ecosystem list
scitex-dev ecosystem audit-all

from scitex_dev import detect_mismatches, verify_versions
mismatches = detect_mismatches()
verify_versions()

Requires Python 3.9+. Designed for use within the SciTeX ecosystem; utility outside that context depends on whether your project structure matches SciTeX conventions.

Verify before relying

• Whether the ~130 audit rules are documented and discoverable outside the source code.
• How the linter plugin system (scitex_dev.linter.plugins entry point) is extended by other packages.
• Whether skills exported to ~/.claude/skills/scitex/ work with Claude Desktop or other MCP clients without additional setup.

Package facts

Evidence: scitex_dev-0.49.3-py3-none-any.whl

Keywords: scitex, developer-tools, documentation, llm-friendly, mcp

Tags

More Build Tools packages

Further reading