agentsclimarketplace

Review severity tiers

Skill bostonaholic/team/skills/review-severity-tiers

Drive a feature from idea to PR with a team of Claude Code agents.

Install
npx -y skills add bostonaholic/team --skill review-severity-tiers

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 8 stars8 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

The authoritative severity-tier map for aggregating reviewer verdicts — gate types by reviewer, the Blocking/Major/Minor tiers with the auto-fix boundary, the consult guard, and the 5-round cap. Load when aggregating review findings, deciding a pipeline gate, or sorting a finding into a severity tier.

SKILL.md

3.1 KB, as published. Nobody here has run it

Review Severity Tiers

Gate Types by Reviewer

ReviewerGate TypeBlocks Ship?
security-reviewerHARDYes — critical or high findings are non-negotiable
verifierHARDYes — tests must pass, build must succeed
code-reviewerHARDYes — blocking issues must be resolved
ux-reviewerAUTO-FIXREQUEST CHANGES is auto-applied in the loop (a major); only COMMENT notes may reach you
technical-writerADVISORYNo — findings recorded, pipeline proceeds

Severity Tiers and the Auto-Fix Boundary

There is no single "blocker/critical/major/minor" scale — reviewers raise findings in three different vocabularies (Conventional Comments issue/suggestion/nitpick, security CRITICAL/HIGH/MEDIUM/LOW, and the APPROVE/REQUEST CHANGES/COMMENT verdict). This table is the authoritative map from any of those onto the action the orchestrator takes. Every finding lands in exactly one tier.

TierFindings in this tierAction
Blockingissue (blocking), code-reviewer REQUEST CHANGES, security CRITICAL/HIGH, any verifier failureAuto-fixed in the loop. Never surfaced to the user.
Majorsuggestion (non-blocking), security MEDIUM, ux-reviewer REQUEST CHANGESAuto-fixed in the loop. Never surfaced to the user.
Minor and belownitpick (non-blocking), security LOW, technical-writer GAPS, any COMMENT-level noteSurfaced to the user for a decision — but only after Blocking and Major are clean.

The consult guard (non-negotiable). While any Blocking or Major finding remains unresolved, the orchestrator MUST NOT present findings to the user or ask which ones to address. It loops the implementer automatically. The user is consulted exclusively for the remaining Minor-and-below findings, and only once the loop has driven Blocking and Major to zero. A consult prompt that lists a blocking or major finding is a defect.

Aggregating Verdicts

When multiple reviewers produce verdicts, aggregate them into a single pipeline gate decision:

  1. If ANY Blocking or Major finding exists -> pipeline gate FAILS — loop back to IMPLEMENT automatically, with no consult.
  2. If only Minor-and-below findings remain -> pipeline gate is CONDITIONAL: present them to the user, who decides. If none remain, proceed to SHIP.
  3. If no findings remain -> pipeline gate PASSES (proceed to SHIP).

The loop continues until Blocking and Major are zero, capped at 5 rounds; at the cap, escalate with the full unresolved-findings summary.

Blocking and Major failures are never aggregated away and never surfaced for triage. A single CRITICAL security finding blocks shipping regardless of how many other reviewers approved.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.