ai-ml/skills-library
SHA256
14
6
Fork 2
Code Issues 6 Pull requests Projects Activity Actions
skills-library/REDHAT.md
Justin Wheeler 2f9c810cda
 🎉 Initial commit: Integrate Agentic SDLC Starter Kit 
Establish the Meeting Secretary repository with foundational files for
an agentic-first development workflow. This includes guiding principles
for teams working with AI coding assistants, a reference for Red Hat's
AI-assisted contribution policy, and documentation on the agentic-first
software development philosophy.

These foundational files provide common guardrails and conventions for
Fedora contributors collaborating on prompt engineering and AI/LLM
tooling in a public, community-driven context.

Assisted-by: Claude Opus 4.6 (1M context)
Signed-off-by: Justin Wheeler <jwheel@redhat.com>
2026-04-15 19:21:41 -04:00

3.2 KiB
Raw Permalink Blame History

Red Hat — AI-assisted development (project reference)

Canonical policy: Guidelines for Responsible Use of AI Code Assistants (supplement to Red Hat's Policy on the Use of AI Technology). If this file conflicts with the wiki, the wiki wins—update this file.

Summary — seven guidelines

  1. Human in the loop: Treat AI output as suggestions; review, test, and modify before integrating.
  2. Approved tools only: Use tools and models approved through Red Hat's AI assessment (and Buy@RH where applicable).
  3. No sensitive data in prompts: Do not input confidential, personal, or proprietary/customer data (passwords, API keys, credentials, PII). Sanitize snippets; prefer synthetic data.
  4. Matching / third-party similarity: If the tool shows "match" information for third-party code, verify before use. AI output may resemble copyrighted training data—investigate license implications when unsure.
  5. Upstream policies: Before contributing to open source, comply with that project's AI rules (some prohibit or restrict AI-assisted contributions).
  6. Mark substantial AI-generated portions: Use commit trailers and/or comments as below.
  7. Copyright and license notices: Follow team practice for file headers; substantial generation with little human creative input especially needs clear marking.

Attribution — commits and PRs

Use one of these in the commit message footer (or clearly in the PR/MR description):

Assisted-by: <name of code assistant> <model name> <model version>
Generated-by: <name of code assistant> <model name> <model version>
  • Assisted-by: You directed the work and edited meaningfully (default for typical use).
  • Generated-by: A substantial portion was generated with minimal human edit (e.g. full file scaffold).

Prefer Assisted-by: or Generated-by: over Co-Authored-By: for AI tools — AI is not a co-author in the traditional sense, and Co-Authored-By: may have CLA/stats implications in some projects.

Source file comments (when needed)

// Generated by <name of code assistant> <model name> <model version>
// Assisted by <name of code assistant> <model name> <model version>

Place at top of file or start of the relevant block, per project conventions.

Quick decision

  • Meaningful human direction and edits → Assisted-by
  • Large generated chunk, little change → Generated-by
  • If unsure → Assisted-by

Proactive warnings (for AI assistants)

  • Secrets / PII / proprietary data: If the user pastes or asks to use them in prompts, warn and suggest redaction or synthetic data.
  • Upstream: If the task mentions kernel or a known restrictive upstream, warn to verify that project's AI contribution policy.
  • Large generated blocks: Remind to mark in commit and optionally in-file comment per the attribution rules above.

EU / works council

Some geographies require works council or local review before adopting tooling. Follow your management chain if unsure.

Upstream always

Check target project policy before submitting AI-assisted patches (kernel and other communities may have strict rules).