Skip to main content
An adjudicated Cortex Bounty report receives one of four verdicts. A valid finding contributes evidence toward a score; it is not a guaranteed reward. The public leaderboard counts reports, while the subnet applies eligibility, precision and severity rules to choose at most one champion per scoring snapshot. The adjudicator decides the finding’s verdict and severity. The published evidence is the input to subnet scoring. Publishing a verdict does not establish that weights have been submitted or a payment has occurred.

​
The four statuses

​
Severity and eligibility

Every newly published valid finding requires a severity: Among participating hotkeys, a contender must satisfy all of these gates:
  • At least three decided reports: valid plus malicious verdicts.
  • Nonnegative net credit: each valid report adds 6, 25, 50 or 100 points by severity; each malicious report subtracts 100.
  • No valid report missing severity.
  • At least 60% precision: valid reports divided by valid plus malicious reports.
  • At most 50% triage noise: duplicate plus already-fixed reports divided by all adjudicated reports.
  • Strictly better precision than the current candidate champion during snapshot selection. A tie does not displace that candidate.
Candidates are considered by descending valid count, with hotkey ordering breaking count ties. The eligible champion’s score combines precision with average severity weight. Other participants receive no positive score for that snapshot, even when they have valid findings. These percentages describe scoring inputs, not a payout rate.

​
What a verdict has to carry

A verdict is refused unless it carries both of these, and each has to be at least 16 characters long:
  • problem found, a statement of what the defect actually was.
  • justification, the reasoning behind the status that was chosen.
The adjudicator recorded on the row is an agent or service identifier, not an email address. A valid verdict also needs severity. A duplicate needs a public related report, with a reference chain ending at a non-duplicate verdict; missing targets, self-links and cycles are refused. The problem statement and justification are public. An optional public-safe summary may also be published after review. The raw report body and chat transcripts remain private.

​
Why a report is not credited

Use this as a checklist when a report has not appeared where you expected it.
A pending report has no public detail or leaderboard credit. It contributes only to the aggregate backlog. A backlog is not a verdict. See Transparency API.
A new valid verdict requires at least 24 characters of plaintext finding. The body stays private; length is only a publication gate, not proof that a finding is substantive.
The problem found and the justification each need at least 16 characters. A verdict that does not meet that is refused rather than published, so the row stays pending.
Historical valid findings without severity are withheld from scoring lists and counts until reviewed again. A retained source permits repricing; the historical verdict alone does not establish scoreable credit.
A duplicate never counts as valid, however good the finding is. Its related report must already be public. The full reference chain must reach a non-duplicate verdict; a chain rooted in an unpriced finding is withheld from scoring lists.
One hotkey belongs to one account. A second account cannot claim a hotkey that is already paired, so a report filed from the wrong account is not credited to that hotkey. See Pair a hotkey.

​
Unavailable evidence is not a zero

The scorer requires one consistent publication revision, complete report pagination, matching aggregate counts and priced valid findings. Missing adjudication, unpriced findings or an initial unworked backlog make the scoring input unavailable. An unsupported stored verdict or severity fails the public read instead of being relabelled as a penalty. An unavailable feed produces no positive Bounty score for that snapshot; it is not evidence that an individual miner performed badly. Check the Transparency API before interpreting an empty leaderboard.

​
How weights follow valid rows

Cortex Bounty accounts for 20% of subnet emission. This is the challenge’s allocation, not a promised amount per report. The account product publishes the evidence; the subnet calculates scores and validators independently verify the allocation before chain submission. A report, leaderboard entry or successful publication does not prove that this full path completed. No fixed payment amount or arrival time is guaranteed by this page.