Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/XxYouDeaDPunKxX/PA-PVP-mini/llms.txt

Use this file to discover all available pages before exploring further.

Severity has exactly three levels. Each level has a specific definition from the protocol — there is no room for subjective escalation. When severity is ambiguous, a tiebreaker rule forces the lower value and requires an explanation, which prevents inflation and keeps severity meaningful across rounds. In finding output lines the middle level is abbreviated as MED; in the definitions below it appears as MEDIUM, matching the protocol’s SEVERITY section.

Severity Levels

Level nameOutput tagDefinitionTypical examples
HIGHHIGHBreaks core function, correctness, execution viability, or downstream trustMissing termination condition, undefined behavior on valid input, output that corrupts downstream consumers, logic that silently produces wrong results
MEDIUMMEDWeakens reliability, robustness, auditability, or prioritization without causing core collapseRetry logic with no cap, missing audit trail for state changes, ambiguous priority ordering between two rules, no error surface for recoverable failures
LOWLOWReal structural weakness with limited blast radiusInconsistent naming convention that could cause confusion in one edge case, a constraint that is technically present but not enforced in one minor branch

Tiebreaker Rule

When the severity of a finding is ambiguous — meaning it could reasonably be assigned either of two adjacent levels — the protocol requires the reviewer to choose the lower severity and explain why. The explanation must be included in the finding output. This rule serves two purposes: it prevents severity inflation, and it makes the reviewer’s reasoning visible. A finding that is upgraded from LOW to HIGH without a clear rationale is a signal that the reviewer is overstating impact.
The tiebreaker rule keeps the reviewer honest. Severity cannot be inflated silently — any time the boundary between two levels is unclear, the lower level is assigned and the reasoning is stated. If you see a HIGH finding and disagree with the severity, the absence of a tiebreaker explanation in that finding is itself a structural weakness you can surface in the next round.

Using Severity for Fix Prioritization

Severity is the primary signal for ordering fixes, but it should not be read in isolation across rounds. Start with HIGH findings. A HIGH finding indicates a problem that breaks core function or correctness. These must be resolved before the artifact can be considered structurally sound. Treat CONFIRMED MED findings seriously. A MED finding that has been confirmed across two or more rounds — meaning the independent reviewer identified it independently and it survived challenge — is a more reliable signal than a single-round HIGH finding. Confirmation across rounds removes the possibility that a finding was overstated or misread in one pass. Do not dismiss LOW findings as noise. LOW findings are real structural weaknesses. They have limited blast radius in isolation, but they can interact with other weaknesses in ways that elevate impact. A CONFIRMED LOW finding that has survived multiple rounds warrants attention even if its standalone severity appears minor. The combination of severity and round label gives the most complete picture. A single-round HIGH that gets CHALLENGED in Round 2 may deserve less immediate action than a CONFIRMED MED that has appeared in every round.

Build docs developers (and LLMs) love