This policy describes how MetaFlowKit decides what to publish, how claims are checked, where AI assistance fits, and what happens when we get something wrong. It is written to be checkable: if the site does not behave the way this page describes, the site is at fault, not the page.
Mission and scope
MetaFlowKit helps people diagnose, estimate and build reliable no-code automations. Our initial platform focus is n8n, Make, and Zapier, and our editorial categories are Diagnostics, Usage Guides, and Comparisons.
We are not a general technology news site, a template marketplace, or a broad software-review publication. We do not attempt to cover every automation tool, and we do not publish on a topic simply because it attracts search traffic. If we cannot say something useful and defensible about a problem, we would rather not have a page about it.
The publication is edited under the collective identity MetaFlowKit Lab Team. Content is written once in global English rather than duplicated into country-targeted variants.
Accuracy standard
Update-sensitive claims must be verified against primary sources before publication. That means the platform’s own current documentation, release notes, status pages, or the product interface as directly observed — not another article, not a forum answer, and not recollection.
Claims we treat as update-sensitive include pricing and plan structures; quota, rate, and concurrency limits; how a platform counts operations, tasks, executions, or credits; node and app capabilities; default settings; and any statement about what a platform “does” as a rule.
We also separate different kinds of statement rather than blending them into a single confident voice:
- Fact. A verifiable claim traceable to a primary source or a direct observation, with the date it was checked.
- Calculation. Arithmetic applied to stated inputs. Inputs are synthetic unless identified otherwise, and the formula is shown.
- Inference. A reasoned conclusion from evidence that is not itself documented. Inferences are marked with language that signals they are reasoning, not confirmation.
- Recommendation. Our judgement about the better option given stated constraints, with the constraints stated.
- Opinion. A view about design, tooling, or trade-offs that another competent practitioner could reasonably reject.
Where evidence is thin or conflicting, we state the uncertainty. Writing “this is unconfirmed” costs a page very little; writing false certainty costs a reader a production incident.
How we use AI
We use AI assistance and say so openly. It can help organise research, structure an outline, produce a first draft, tighten unclear writing, review code and expressions for errors, and run consistency checks across a draft before an editor reads it.
AI is not a source. An AI system cannot confirm what a platform currently charges, what a setting currently does, whether a limit still applies, or whether a workflow ran successfully. It can produce fluent text that is confidently wrong, and it does so most convincingly in exactly the areas we cover — version-specific behaviour, pricing details, and interface labels.
So the rule is straightforward. Anything AI assistance contributes is treated as draft material requiring verification, never as evidence. Update-sensitive claims are checked against primary sources by a human editor. Interface labels and setting names are checked against the product or its documentation. Code and expressions are reviewed by a person before publication. Accountability for every published claim rests with the editorial team, not with a tool.
We do not publish AI-generated screenshots, AI-generated interface mock-ups presented as real, or AI-produced citations that have not been opened and confirmed.
Claims about testing
We never claim a test that was not run. This is the single rule most easily broken by careless wording, so it is stated plainly.
Where a workflow, fix, or configuration has genuinely been run, the page records the environment and versions used, the prerequisites, the inputs, the expected output, and how the result was verified. Where a test was partial, the page describes what was covered and what was reasoned about rather than implying full coverage.
Where no test was run, the page rests on documentation and reported symptoms and says so. Phrases like “we tested”, “in our testing”, and “we confirmed” are reserved for work that actually happened. The full method, including our three evidence labels, is set out in How We Test.
Corrections and updates
Platforms change, and some of what we publish will become wrong. Our commitments:
- Material errors are fixed. A factual error affecting the outcome of a page is corrected once identified and confirmed, rather than quietly softened.
- Meaningful changes are clarified. Where a correction changes what a reader should do, the page notes what changed rather than leaving a silently altered instruction.
- Dates stay honest. Genuine publication and update dates are preserved. We do not refresh dates to make pages look current when no substantive work was done.
- Significant revisions are noted. Where a page has been substantially rewritten because a platform changed, that is recorded in a way a reader can see.
Minor copy edits, formatting fixes, and typographical corrections do not require a note. The test is whether a reasonable reader would act differently because of the change.
Originality
Every page is written for this site. We do not republish or lightly rewrite other publications’ articles, spin existing text into near-duplicates, or generate near-identical pages from a template with the platform name swapped.
Screenshots are original and taken from actual use, with credentials, tokens, endpoint URLs, account identifiers, and personal data blurred. We do not fabricate screenshots or reuse a vendor’s marketing imagery as though it were our own observation. We do not invent quotes, attribute statements to people or companies that did not make them, publish fabricated statistics or readership figures, or display trust badges we have not earned.
Where we draw on someone else’s work, we say so and link to it directly, as described in the Source Policy.
Commercial content
Current commercial status of this site:
- Affiliate relationships: None currently
- Advertising or ad-management services, including Grow: None currently
Those values are set during editorial configuration and describe what actually exists, not what is planned. We do not describe a commercial arrangement as active before it is.
The rules that apply if and when such relationships exist:
- Compensation does not determine a technical conclusion. Whether a platform handles retries well, or a plan’s limits suit a workload, is decided by evidence and stated constraints.
- Required disclosures appear where a reader encounters the link or placement, not only in a policy page they may never open.
- Advertising placements are visually distinguishable from editorial content.
- Advertisers and affiliate partners do not receive advance review of, or approval over, editorial conclusions about their products.
- An arrangement that would require softening an accurate finding is declined.
We are not claiming that commercial pressure is impossible — that would be a claim we could not support. We are describing the separation we operate and the disclosures that make it checkable.
Conflicts, gifts, and contributions
Where anything of material value could reasonably be thought to influence coverage — free or extended plan access provided for review, gifted services, paid consulting with a platform we cover, or a financial interest in a vendor — it is disclosed on the relevant page.
Review access provided by a vendor is stated as such, and does not come with editorial conditions. Where a vendor declines to provide access, we say that rather than implying we chose not to look.
External contributions are not currently invited through an open contributor programme, and we do not describe one that does not exist. If external contributions are published in future, the contributor’s relationship to the platforms covered will be disclosed on the page and the same verification standards will apply.
Feedback and appeals
If you believe a page is wrong, incomplete, or unfair to a product, tell us. The most useful reports include the page URL, the specific statement in question, the current primary source that contradicts it, and the relevant date or version. Reach the editorial team through the Contact page.
We will correct what is wrong. Where we respond to a substantive report and disagree, we will explain why. Public article comments are currently disabled, so email is the route for this.