granite-guide
Build with IBM's Granite models — enterprise open models for code, language, and business applications.
- License 许可
- MIT
- 最近更新
- 2026-11-07
- Stars
- ★ 0
- Forks
- 1
01 前言 Front Matter
- name
- granite-guide
- category
- ai-research
- description
- Build with IBM's Granite models — enterprise open models for code, language, and business applications.
03 SKILL.md 正文
Overview
Granite (IBM) is an enterprise-oriented open model family spanning code, language, and specialized business models (including time-series and vision variants) — built with IBM's enterprise focus: training-data transparency, indemnification offerings, and watsonx platform integration. The positioning is "open models safe for enterprise" — Apache 2.0 licensing on key releases plus IBM's enterprise assurances.
For builders in IBM-centric enterprises, Granite is the natural open-model choice: watsonx integration, enterprise support, and licensing clarity. For others, Granite deserves evaluation as a solid open family — particularly where licensing simplicity (Apache 2.0) and enterprise assurances matter.
The distinctive note: IBM's indemnification and transparency commitments address enterprise legal concerns directly. Evaluate whether that assurance has value for your organization's risk posture.
When to use
- IBM/watsonx-centric enterprises building on open models.
- Applications needing Apache 2.0 licensed open models.
- Enterprise deployments valuing vendor indemnification.
- Code and language tasks on open models with enterprise support.
- Regulated industries needing documented, assured AI components.
- Time-series and specialized enterprise tasks (Granite's specialized variants).
Core concepts
- Enterprise open models: open weights with enterprise assurances — the combination is the differentiator.
- Apache 2.0 licensing: permissive licensing on key releases. Verify per release; the simplicity vs. community licenses is real.
- Indemnification: IBM's IP indemnification offerings for enterprise deployments. Understand the scope — what it covers and what it doesn't.
- Code and language families: Granite code models and language models for respective tasks. Benchmark each on your tasks.
- Specialized variants: time-series (Chronos-style), vision, and other enterprise-task models. Match variant to task.
- watsonx integration: native platform integration for IBM-stack teams — deployment, governance, and tooling.
- Training transparency: documented training approaches supporting enterprise risk assessment.
- Enterprise support: vendor support for open models — unusual and valuable for risk-averse organizations.
Practical workflow
- Clarify assurance requirements. What does your organization need: licensing simplicity, indemnification, support, transparency? Get specific.
- Benchmark model quality. Granite code and language models vs. alternatives (Llama, Qwen, DeepSeek) on your tasks. Assurances don't replace quality evaluation.
- Verify licensing per release. Confirm Apache 2.0 (or other) terms for your chosen models and use case.
- Understand indemnification scope. Read the actual terms — coverage, exclusions, and claim processes.
- Evaluate watsonx fit. If on IBM stack: test the integrated deployment and governance paths. If not: evaluate Granite as standalone open models.
- Test specialized variants. For time-series or other specialized tasks: benchmark the Granite specialist against alternatives.
- Pin, monitor, document. Version pins; quality monitoring; the assurance documentation filed for compliance.
Checklist for Granite in production:
- Assurance requirements specified and matched to IBM's offerings.
- Quality benchmarked against open-model alternatives.
- License verified per release.
- Indemnification scope understood.
- Specialized variants tested where relevant.
Common pitfalls
- Assurance as quality proxy. Enterprise assurances address legal risk, not model quality. Benchmark anyway.
- Indemnification over-read. Assuming broader coverage than the terms provide. Read the scope.
- No peer benchmarks. Choosing on enterprise comfort without testing against Llama/Qwen/DeepSeek on your tasks.
- watsonx assumed. Committing to the platform without evaluating whether the integration benefits your team.
- License per-release unchecked. Assuming all Granite releases share terms.
- Specialized variants ignored. Missing the time-series and other specialists that may fit enterprise tasks well.
- Support expectations vague. Not clarifying what enterprise support actually covers before depending on it.
- Static enterprise choice. Deciding once on assurance grounds and never re-benchmarking quality as the open field advances.
04 同仓库更多 Skills
web-scrape
Turn web pages into LLM-ready markdown via the Firecrawl API (renders JS, anti-bot handling). Use when browser_open/fetch fails on JS-heavy pages, or when you need clean markdown/structured extraction from a site. Requires FIRECRAWL_API_KEY.
design-taste-frontend
tasteskill v2, the anti-slop frontend skill. Art direction for websites, distilled from a study of 247 curated gallery sites (One Page Love, Awwwards Sites of the Day). Deals researched directions (idea, composition, type, palette, photographs, section blueprints) from a library, enforces a strict no-slop house style with a real-browser inspector, and runs a look-and-fix loop on screenshots. Use this whenever you build, design, redesign, restyle, polish or review a landing page, marketing site, portfolio, product page, one-pager, hero or any public-facing web page, in any stack (HTML/CSS, React, Next.js, Vue, Svelte, Astro, Tailwind) — also when the user says a page looks generic, templated, "AI-made", sloppy, or asks to make it look designed, premium, award-level or less like every other site. Not for dashboards or app screens.
test-driven-development
Use when implementing any feature or bugfix, before writing implementation code
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes
requesting-code-review
Use when completing tasks, implementing major features, or before merging to verify work meets requirements
brainstorming
You MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation).