sdd-quality-loop

Plugin sdd-quality-loop from marketplace sdd-plugins in aharada54914/sdd-forge. Commands read from the repo's manifest — 5 days ago.

Skills (6)

• cross-model-verify

  Orchestrate blind parallel cross-model panelist verification for a critical SDD task. Runs prepare-panelist-input (consent + sanitize), invokes panelists blind in parallel (Claude via Agent tool; GPT/Gemini via CLI runners), collects verdict JSONs, then prompts the user to run check-cross-model to g

• fix-by-review-ticket

  Apply only a human-approved repository review-ticket YAML fix, run scoped checks, and return the task to Implementation Complete for another quality gate.

• quality-gate

  Independently verify an Implementation Complete task with a Default-FAIL verification contract, deterministic checks, and an isolated critical reviewer, then decide whether it is Done.

• sdd-sudo

  Explain human-owned sudo activation, status and revocation for routine approval waiting; never issue tokens through agent tools.

• wfi-audit-cycle

  Orchestrator for the WFI proposal audit. Runs 2 sequential independent audit cycles on a WFI-NNN.md Draft before presenting it to a human for Approved. Cycle 1 (wfi-auditor-a) audits proposal quality. Cycle 2 (wfi-auditor-b) audits impact and risk. Each auditor runs as a fresh isolated agent. The or

• workflow-retrospective

  Measure how the SDD workflow itself performed (rework cycles, blocked tasks, review tickets, quality-gate failures) and propose human-approved improvements to project workflow files.

Subagents (5)

• sdd-evaluator

  Independent skeptical evaluator for SDD quality gates. Reviews one Implementation Complete task against the approved specification in a fresh context. Read-only; returns PASS or NEEDS_WORK with classified findings.

• sdd-panelist-gemini

  Blind independent panelist for SDD cross-model verification. Reviews one sanitized input bundle and returns a cross-model-verdict/v1 JSON. Read-only; never writes code or approves tasks. Represents the Google/Gemini vendor slot.

• sdd-panelist-gpt

  Blind independent panelist for SDD cross-model verification. Reviews one sanitized input bundle and returns a cross-model-verdict/v1 JSON. Read-only; never writes code or approves tasks. Represents the OpenAI/GPT vendor slot.

• wfi-auditor-a

  WFI Proposal Quality Auditor for audit cycle 1. Reviews WFI-NNN.md for evidence quality, root cause plausibility, why-why (5 Whys) chain validity, category-appropriate language, concrete proposed changes, and measurable expected effects. Read-only; returns PASS, NEEDS_REVISION, or BLOCKED with class

• wfi-auditor-b

  WFI Impact and Risk Auditor for audit cycle 2. Reviews the revised WFI-NNN.md for verification plan quality, change scope proportionality, unintended consequences, implementation feasibility, and language compliance (second pass). Read-only; returns PASS, NEEDS_REVISION, or BLOCKED with classified f

Hooks (2)

• PreToolUse

  sh "${CLAUDE_PLUGIN_ROOT}/scripts/kill-switch.sh"

• PreToolUse

  sh "${CLAUDE_PLUGIN_ROOT}/scripts/sdd-hook-guard.sh" --emit exit

Similar plugins