Devnet prototype · scores are indicative
FlopRadar: independent trust scores for Flop Network apps.
We find the apps, tools and guides built on the Flop Network, test them, check whether they do what they claim, and flag the scams. Nobody pays for a better score.
- Apps found
- 53
- Ranked overall
- 40 +10 guides ranked in category
- Scam flags
- 3 excluded from every ranking
- Does it do what it claims?
- Verified 42
- Partly 7
- Not verified 3
- Untestable 1
01 · Leaderboard
Trust leaderboard
Scored 0–100 on reliability, quality, safety, speed and cost, then checked against what each project claims. Select a row for evidence, safety flags and the claims note.
Everything except guides, ranked together. Guides are ranked within their category in the Guides tab.
| Loading the latest scores… | |||||||||
Download leaderboard.json Free, the same data this page reads. Updated 2026-11-10.
02 · Fake-token checker
Is this “FLOP” token real?
No. No official FLOP token exists before mainnet/TGE, targeted for Q1 2027. Paste an address or a token name to see whether it is a known fake. The check runs in your browser; nothing is sent anywhere.
03 · Scam alerts
Excluded projects
These projects tripped a high-severity safety flag, so they are left out of every ranking. Be careful with them.
04 · Methodology
How a score is made
An independent judging agent finds Flop Network apps, tests them and scores each 0–100. The method is fixed and public. Payment never changes a score.
Five measured metrics
- Reliability25
- Quality25
- Safety25
- Speed15
- Cost10
Weights out of 100. Cost only counts for apps that buy inference. A metric that applies but could not be measured enters as a neutral 50 at half weight.
What each metric means
- Reliability
- On-chain settle rate for inference apps; live-URL uptime plus repo freshness for tools and dashboards.
- Quality
- A fixed 8-prompt test set graded by rules and a judging pass for inference apps; documentation depth and usefulness for tools.
- Safety
- Scan for token-sale promotion, seed-phrase or key requests, fake “Official” claims, Sybil evasion and lookalike domains. Below 50, the whole score is capped at the safety score.
- Speed
- Time to first token and full response, or HTTP response time of the live site.
- Cost
- Devnet FLOP per answer from the settled receipt, on a log scale.
Does it do what it claims?
For every item we list 2–4 concrete claims from its README or homepage and test what we safely can: live URLs and APIs, static repo checks, pinned digests. Repo code is never executed.
| Verdict | Quality penalty |
|---|---|
| Verified | 0 |
| Partly | −5 |
| Not verified | −15 |
| Untestable | 0 |
Community votes, capped
Votes can add at most 15% of the final score. The community score is the Wilson lower bound of the weighted upvote share, so a handful of votes barely moves anything, and the weight only reaches 15% at 50 effective weighted votes.
trust = (1 − w) · measured + w · community w = 0.15 · min(1, votes / 50)
Voter weights: anonymous 0.2, wallet-signed 1, wallet with settled sessions 3. Self-votes and voting blocs are discarded. The safety gate runs first, so votes can never lift a flagged project back into the rankings.
Known limits. Devnet only: devnet FLOP has no value and the network may reset. Probes are a single snapshot, the claims check is a one-time manual pass, and live inference tests for third-party apps are queued until their endpoints are public. Treat scores as indicative.
05 · API Coming with testnet
An API for agents
Agents will be able to query scores and run fresh checks, paying in FLOP over x402 (HTTP 402). The leaderboard stays free. Paying buys speed and depth, never a better score; owner-requested audits use the same method and are labelled as such.
| Endpoint | Returns | Price (placeholder) |
|---|---|---|
GET /v1/apps | Full leaderboard JSON | Free, rate-limited |
GET /v1/scam-check?token= | Is this “FLOP” token fake? | Free |
GET /v1/check?app= | Trust score, claims verdict, safety flags, last tested | 0.05 FLOP |
POST /v1/check/live | Fresh endpoint probe plus 3 judged inference calls | 1 FLOP |
POST /v1/audit | Full audit of a new app or repo: claims, safety, about 16 judged calls, listed on this page, webhook when done (about 30 min) | 10 FLOP |
Download leaderboard.json Available now, free: the full dataset behind this page.
06 · Community
Community voting opens soon.
People and agents will be able to up- or down-vote apps, with wallet-signed votes from accounts with real settled sessions counting most. Votes stay a small, capped input; the tests decide most of every score.