Editorial Policy

1. Our Editorial Standards

Every page on AI Code Toolkit is written and reviewed against the same core standards, regardless of who authored it:

• Accurate. Every technical claim is verifiable against primary sources (official Anthropic docs, tool documentation, our own testing).
• Tested. Every code snippet runs. If we haven't run it, we don't publish it.
• Current. Content reflects the version of the tool that shipped most recently. Where behavior differs between versions, we say so.
• Honest. When Claude Code isn't the best fit for a use case, we say that plainly. When we're uncertain, we say that too.
• Original. We do not republish content from other sources without permission and full attribution. Our writing is our writing.
• Useful. Every page answers a real developer question we've seen asked repeatedly, not just a keyword we want to rank for.

2. The Testing Process

Every code-heavy page — slash commands, subagent configs, hooks, MCP integrations, CLAUDE.md templates — goes through a testing gate before publishing:

1. The author writes the snippet against a real Claude Code project.
2. The author documents the specific Claude Code / SDK / Anthropic API version used and the date tested.
3. A separate editor runs the snippet independently on a different environment to confirm it works.
4. The page is marked "Last tested: [date]" and, where relevant, "Anthropic Claude Code v[X.Y.Z]".
5. If the snippet involves paid API calls, we log the approximate cost in the page for reader transparency.

If a snippet stops working — for example, because Anthropic changed a parameter default — we treat that as a P0 correction and ship a fix as soon as we detect it, ideally within 48 hours.

3. Bylines & Author Attribution

Every content page carries a byline from one of our senior editors. Byline rotation is not random — each byline corresponds to a real editor who reviewed and stands behind the page:

• Ahmed R. (initials AR) — senior editor. Backend, DevOps, database, security. Reviews and bylines odd-numbered snippet pages.
• Sana K. (initials SK) — senior editor. Frontend, AI integration, prompt engineering, TypeScript. Reviews and bylines even-numbered snippet pages.
• Guest bylines. Occasional invited experts (Anthropic engineers, dev-tool founders, community MVPs). Always identified clearly with the guest's affiliation.

When multiple authors contribute meaningfully, both bylines appear. When AI assistance is used in drafting (see Section 8), the byline still identifies the human editor responsible.

4. Correction Policy

We take corrections seriously. When we find (or you point out) an error, our process is:

1. Fix the error immediately or, if the fix requires research, add a visible caveat while we work.
2. Add a "Corrections" note at the bottom of the affected page with the date and a summary of what changed.
3. For material errors that may have affected reader decisions (e.g., a wrong pricing figure, a broken security assumption), we may email newsletter subscribers.
4. For factual errors we did not catch before publishing, we thank the reader who reported it publicly (with permission) and privately.

We distinguish between corrections (facts that were wrong when published) and updates (facts that became wrong because the world changed — e.g., Anthropic released a new version). Both are noted with dates; only corrections carry the "corrected on" label.

5. Sources & Attribution

We rely on primary sources whenever possible:

• Anthropic's official documentation for Claude Code features and API surface.
• Tool vendor documentation for third-party MCP servers, frameworks, and dependencies.
• Our own testing for anything we can verify hands-on.
• Anthropic release notes and changelog entries for feature timing and version specifics.

Secondary sources (blog posts, community discussions, Stack Overflow) are used only for context or to identify problems worth writing about — not as authoritative sources.

When we quote or reference another source, we credit it with a link. We do not paraphrase closely enough to constitute republishing.

6. Sponsored Content Policy

From Month 6 onward we may accept sponsored content. When we do, we follow strict rules:

• Clearly labelled. Every sponsored page carries a "Sponsored by [brand]" banner at the top, above the fold.
• Editorially independent. The sponsor does not review or approve editorial content on non-sponsored pages.
• Fact-checked to the same standard. Sponsored pages go through the same testing and accuracy gates as editorial content.
• No sponsored comparisons. Our comparison pages are never sponsored. That would defeat their whole purpose.
• No sponsored search rankings. Sponsored pages use rel="sponsored" and are not designed to manipulate our internal ranking or search visibility.

Affiliate content is different from sponsored content — see our Affiliate Disclosure.

7. Community Submissions

Developers can submit their own slash commands, hooks, subagent configs, and MCP integrations via our GitHub repo. Community submissions are reviewed under the same standards as staff-written content:

1. Submitter opens a pull request with the snippet plus a brief write-up.
2. Editor tests the snippet independently.
3. Editor works with submitter on any needed revisions.
4. Once approved, the page is published with the submitter's byline (and a link to their GitHub or website if provided).
5. Contributors are eligible for our upcoming community marketplace revenue share (Q4 2026), initially 70/30 in favor of the contributor.

Full contribution guidelines live in the GitHub repo README.

8. AI-Assisted Content Disclosure

We use AI tools — including Claude — as writing and research assistants in the editorial process. This is unavoidable and fine, but transparency matters:

• AI is used to draft initial outlines, suggest examples, and cross-check facts against known sources.
• Every page is reviewed, revised, and approved by a human editor who takes editorial responsibility.
• Code snippets are never taken directly from AI output without independent testing.
• Bylines identify human editors, not AI models. We do not publish AI-only pages.
• Where AI-generated illustrations or images are used, we mention it in the image caption.

Our editorial standard is: AI is a tool. Judgment is human.

9. Freshness Commitment

The AI coding ecosystem changes fast. We meet that with a defined freshness cadence:

• Anthropic Claude Code releases: affected pages updated within 48 hours.
• Third-party MCP server updates: affected pages updated within 7 days.
• Rolling refresh: the full 520-page library is re-reviewed every 90 days.
• Pricing pages: reviewed monthly, updated immediately when a vendor changes prices.

Every page shows a "Last updated" date. If a page is older than 90 days without update, it's automatically flagged in our editorial dashboard for review.

10. Conflicts of Interest

Editors and contributors disclose material conflicts of interest to the editor-in-chief. If we ever review a product from a company where a team member has a direct financial interest (equity, board seat, active consulting engagement), we disclose it in the affected page.

Small affiliate relationships are covered by our public Affiliate Disclosure and are not considered material conflicts.

11. How to Submit a Correction or Feedback

The fastest way to reach the editorial team:

• Email editorial@aicodetoolkit.com with the page URL and a short description.
• Open a GitHub issue on our public repo (if the correction is code-related).
• Use the contact form on the contact page, selecting "Editorial" as the topic.

We aim to acknowledge editorial feedback within 24 hours and ship fixes within 3 business days — often the same day.