Every pa_pvp mini finding is assigned a severity level that indicates how seriously the identified problem affects the artifact’s function, reliability, or trustworthiness. Severity is not about impact on users — it is about impact on the artifact’s structural integrity. A HIGH finding does not mean users will immediately suffer; it means the artifact itself cannot do what it claims to do when the problem is present.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 levels
| Level | Output label | Meaning |
|---|---|---|
| High | HIGH | Breaks core function, correctness, execution viability, or downstream trust. The artifact cannot be relied on for its intended purpose with this problem present. |
| Medium | MED | Weakens reliability, robustness, auditability, or prioritization without causing complete core collapse. The artifact still operates, but it does so inconsistently, unsafely, or in ways that cannot be audited. |
| Low | LOW | A real structural weakness with a limited blast radius. The artifact functions in most cases, but a specific condition or edge case exposes a gap. |
Choosing between adjacent levels
When the severity of a finding is ambiguous — for example, when a problem could be either MED or HIGH depending on interpretation — the reviewer must choose the lower severity and explain why. Example: An artifact contains a constraint that is stated but never enforced. In most executions the constraint is followed by convention, but there is no mechanism that guarantees it. This could be HIGH (if the constraint governs core correctness) or MED (if it governs audit quality but not execution viability). If the artifact still executes correctly when the constraint is violated, the reviewer must assign MED and state: “assigning MED because execution is not blocked by this gap; core function is preserved but auditability is compromised.”The tiebreaking rule toward the lower severity is not about minimizing the problem. It is about intellectual honesty. Overstating severity weakens the credibility of the entire output. If a finding is genuinely HIGH, the evidence will support it without ambiguity.
What severity is NOT
Severity in pa_pvp mini has a specific meaning. It is not a measure of any of the following:- User impact. A problem that would frustrate users is not automatically HIGH. Severity tracks structural damage to the artifact, not downstream experience.
- Urgency. There is no time dimension in severity. A LOW finding in a deployed artifact may need attention faster than a HIGH finding in a draft.
- Effort to fix. A trivial one-line fix can address a HIGH finding. A complex refactor may address a LOW one. Effort and severity are independent.
- Importance to the reviewer. Severity must follow the definitions, not the reviewer’s judgment about what matters most.
HIGH findings
HIGH is assigned when the problem causes the artifact to break at its core: a step that always fails, a constraint that is never enforced despite being load-bearing, or a decision rule that has no coverage for a key case and leaves execution undefined. Examples of HIGH findings:- A decision rule with three branches, where one branch is never reachable because a prior condition prevents it — leaving a class of inputs unhandled.
- A validation step that references a variable that is not defined anywhere in the artifact’s scope.
- A constraint that is declared but has no enforcement mechanism, and the artifact’s correctness depends on that constraint holding.
- A flow where step 4 requires output from step 6, creating an unresolvable dependency cycle.
MED findings
MED is assigned when the problem weakens how reliably or auditably the artifact performs, without preventing it from executing in common cases. The artifact runs, but it runs inconsistently, ambiguously, or in ways that are difficult to verify. Examples of MED findings:- A step that sometimes fails under specific conditions that are not documented or handled.
- A constraint that is stated ambiguously enough that two reviewers would enforce it differently, creating audit gaps.
- A prioritization rule that produces inconsistent results when two conditions apply simultaneously, with no tiebreaker defined.
- A branching rule that works correctly but relies entirely on judgment to apply, with no defined criteria.
LOW findings
LOW is assigned when the problem is real and structural but its blast radius is limited. The artifact works correctly in the vast majority of cases, but a specific edge condition, labeling gap, or fragile assumption is present. Examples of LOW findings:- A label that applies to two different concepts in the same artifact, which could cause misreading under specific conditions.
- An optional step that has no fallback documented for the case where it is skipped.
- A section that is structurally complete but depends on an assumption that is unstated and could fail under edge inputs.
- A rule that works correctly but is phrased in a way that makes it easy to misapply without noticing.