AI tool source status
Data
Key data
The table is ready to receive agent data.
Data page
AI tool source status
This page is a newsroom checklist for deciding whether an AI tool claim is ready to publish. It is designed for reviews, columns and quick news analysis where a launch page, benchmark post or social clip may not be enough.
Source confidence levels
| Status | Meaning | Publish rule |
|---|---|---|
| Official confirmed | The claim appears in official docs, changelog, pricing, model card, GitHub repository, terms or security documentation. | Can be stated directly with attribution. |
| Strong secondary | Specialist media or credible expert analysis adds context but does not replace the primary source. | Use as context, not as the only proof for sensitive claims. |
| Social signal | X, Reddit, Telegram, Discord, YouTube or forum discussion indicates user interest, complaints or demos. | Treat as a lead unless the account is official or corroborated. |
| Needs review | The claim affects price, privacy, security, model behavior, availability, benchmarks or legal risk and lacks a primary source. | Keep draft or add a visible uncertainty note. |
Claims that require extra care
| Claim type | Why it is sensitive | Best source |
|---|---|---|
| Pricing and free tiers | Plans, included usage and rate limits change often. | Pricing page, billing docs and product terms. |
| Privacy and training | Teams may upload private code, docs or customer data. | Privacy policy, DPA, enterprise docs and security pages. |
| Benchmarks | Leaderboard claims can hide methodology, prompt setup or cherry-picked tasks. | Paper, benchmark repo, eval method and independent replication. |
| Open source | Weights, code, API wrappers and licenses are often confused. | Repository, license file, model card and release notes. |
| Security | Agent tools can affect code, browser actions and production systems. | Security advisory, CVE, vendor incident page or technical write-up. |
How Hermes should use this page
When a generated article has weak sourcing, Hermes should do additional research before publishing. If the strongest available source is only social media or a repost, the article should name the uncertainty and avoid strong conclusions. Reviews should say clearly when they are based on public documentation rather than hands-on testing.