qrspi

Plugin qrspi from marketplace qrspi-local in dfrysinger/qrspi-plus. Commands read from the repo's manifest — 1 month ago.

Skills (17)

• design

  Use when research/summary.md is approved and the QRSPI pipeline needs an architecture — proposes approaches, surfaces key architectural decisions with rationale, and defines per-goal acceptance through interactive design discussion

• goals

  Use when starting a new QRSPI pipeline run — captures user intent and constraints through interactive dialogue, then synthesizes a problem-framed goals.md

• implement

  Per-phase implementation orchestrator. In full pipeline mode, resolves symbolic bases from parallelization.md to concrete commits, creates worktrees and stage commits, runs baseline tests, dispatches implementer + reviewer subagents per task per the wave schedule, runs the fix loop, presents the bat

• implementer-protocol

  Cross-cutting QRSPI implementer protocol — dispatch shape, mode payloads, status reporting, ID hygiene, and the BLOCKED escape hatch shared by every implementer subagent.

• integrate

  Use when all current-phase tasks are implemented — merges task branches, runs cross-task integration review, security review, and CI gate

• parallelize

  Use when plan.md is approved and the QRSPI pipeline needs a parallelization plan — analyzes task dependencies and file overlap, decides execution mode, produces parallelization.md with a symbolic branch map; hands off to Implement

• phasing

  Use when design.md is approved and the QRSPI pipeline needs vertical slice authoring, phase boundary decisions, roadmap.md authoring, current-phase pruning, and goal-ID consistency validation — sits between Design and Structure

• plan

  Use when prior artifacts are approved and the QRSPI pipeline needs detailed task specs — breaks structure into ordered tasks with test expectations, dependencies, and LOC estimates (full pipeline requires design+structure; quick fix requires only goals+research)

• prompt-prose-reviewer

  Apply prompt-design rules when reviewing prompt-prose subjects in a diff. Detects which files (or sub-blocks) are prompt prose, applies R1-R7 + cross-cutting principles + finding-type gate, and emits findings with proper change_type tagging. Preloaded by reviewer agents that may encounter prompt pro

• prompt-prose-writer

  Apply prompt-design rules when authoring or planning prompt-prose deliverables. Detects whether a deliverable IS prompt prose, and only then Reads the rules and applies R1-R7 before drafting. Preloaded by agent files that may author prompt prose.

• questions

  Use when goals.md is approved and the QRSPI pipeline needs research questions generated — produces tagged questions that guide the Research step without leaking goals

• replan

  Use between phases when Test signals more phases remain — analyzes completed phase, proposes task updates with severity classification, handles minor updates or major backward loops

• research-isolation

  Pre-Flight check shared by research-phase agents (specialist, collator, reviewer) that refuses dispatches whose prompt contains goals-content or out-of-scope question content.

• research

  Use when questions.md is approved and the QRSPI pipeline needs objective codebase and web research — dispatches parallel specialist subagents per question, collates per-question findings into research/summary.md

• reviewer-protocol

  Cross-cutting QRSPI reviewer protocol — finding schema, change-type classifier, untrusted-data handling, and dispatch contract. Per-channel emission contracts live in sibling files first-party-emission.md and third-party-emission.md.

• structure

  Use when design.md is approved and the QRSPI pipeline needs file/component mapping — maps vertical slices to specific files, interfaces, and component boundaries

• using-qrspi

  Use when starting any conversation — establishes the QRSPI pipeline for agentic software development, requiring structured progression through Goals, Questions, Research, Design, Phasing, Structure, Plan, Parallelize, Implement, Integrate, Test, with Replan firing between phases

Subagents (42)

• qrspi-code-quality-reviewer

  Verifies the implementation is clean, well-structured, and maintainable. Used in both Implement phase (per-task code review) and Test phase (test code review). Runs after spec-reviewer passes.

• qrspi-code-simplifier

  Identifies opportunities to simplify code while preserving all functionality. Deep mode only. Runs after all correctness reviewers pass. Simplifications are suggestions, not blocking.

• qrspi-design-reviewer

  Reviews design.md for artifact-specific quality (correctness, clarity, completeness) per the QRSPI reviewer protocol. Scope/boundary review is handled by qrspi-design-scope-reviewer.

• qrspi-design-scope-reviewer

  Scope/boundary review for design.md. Reads skills/design/owns-defers.md and applies the 3-check scope procedure. Companion to qrspi-design-reviewer (which handles artifact quality).

• qrspi-finding-verifier

  Score a single reviewer finding 0–100 against the /code-review confidence rubric. Read the per-finding file + artifact + lazy-Read upstreams; Write a sidecar score file; return a brief <reviewer_tag>.<finding_id>: <score> line.

• qrspi-goal-traceability-reviewer

  Verifies an unbroken traceability chain from goals through specs through tests to implementation. Deep mode only. Also reused by the Test phase to review generated test code. Runs after all correctness reviewers pass.

• qrspi-goals-reviewer

  Reviews goals.md for artifact-specific quality (correctness, clarity, completeness) per the QRSPI reviewer protocol. Scope/boundary review is handled by qrspi-goals-scope-reviewer.

• qrspi-goals-scope-reviewer

  Scope/boundary review for goals.md. Reads skills/goals/owns-defers.md and applies the 3-check scope procedure. Companion to qrspi-goals-reviewer (which handles artifact quality).

• qrspi-implement-gate-reviewer

  Cross-task batch-gate reviewer dispatched when the user selects "Re-run all reviews" at the Implement batch gate. Reviews the combined wave of task code, specs, and test results for cross-task patterns and gate-level issues.

• qrspi-implementer-lightweight

  Per-task non-TDD implementation subagent for prose / prompt / doc / config tasks (task_type=lightweight). Single-pass implement, no test scaffolding. Per-task model selection is handled by the dispatcher via per-invocation override.

• qrspi-implementer

  Per-task TDD implementation subagent. Handles initial implementation (mode: implement) and fix cycles (mode: fix). Per-task model selection is handled by the dispatcher via per-invocation override.

• qrspi-integration-reviewer

  Reviews merged code from multiple implementation tasks for cross-task integration issues. Dispatched from the Integrate phase after all tasks are merged.

• qrspi-parallelize-reviewer

  Reviews parallelization.md for artifact-specific quality (correctness, clarity, completeness) per the QRSPI reviewer protocol. Scope/boundary review is handled by qrspi-parallelize-scope-reviewer.

• qrspi-parallelize-scope-reviewer

  Scope/boundary review for parallelization.md. Reads skills/parallelize/owns-defers.md and applies the 3-check scope procedure. Companion to qrspi-parallelize-reviewer (which handles artifact quality).

• qrspi-phasing-reviewer

  Reviews phasing.md for artifact-specific quality (correctness, clarity, completeness) per the QRSPI reviewer protocol. Scope/boundary review is handled by qrspi-phasing-scope-reviewer.

• qrspi-phasing-scope-reviewer

  Scope/boundary review for phasing.md. Reads skills/phasing/owns-defers.md and applies the 3-check scope procedure. Companion to qrspi-phasing-reviewer (which handles artifact quality).

• qrspi-plan-apply-fix

  Apply the round's accepted reviewer findings to plan.md and per-task spec files. Reads skills/plan/owns-defers.md, design.md, and structure.md, applies the upstream-contract pre-flight before each finding, and surfaces design/structure-amendment-required findings as Author Notes instead of self-appl

• qrspi-plan-goal-traceability-reviewer

  Verifies bidirectional traceability between goals and plan tasks — every goal traces forward to plan-authored test expectations, and every task traces back to a goal or research finding. Reviews the plan artifact, not task implementations. Runs always (quick + full pipeline).

• qrspi-plan-reviewer

  Reviews plan.md for artifact-specific quality (correctness, clarity, completeness) per the QRSPI reviewer protocol. Scope/boundary review is handled by qrspi-plan-scope-reviewer.

• qrspi-plan-scope-reviewer

  Scope/boundary review for plan.md. Reads skills/plan/owns-defers.md and applies the 3-check scope procedure. Companion to qrspi-plan-reviewer (which handles artifact quality).

• qrspi-plan-security-reviewer

  Identifies fail-open conditions, missing validation, auth gaps, and insecure defaults in the planned design before implementation begins. Reviews the plan artifact, not task implementations. Runs always (quick + full pipeline).

• qrspi-plan-silent-failure-hunter

  Identifies planned behaviors that would swallow errors, silently fall back, leave partial state, or log-and-continue when they should fail loudly. Reviews the plan artifact, not task implementations. Runs always (quick + full pipeline).

• qrspi-plan-spec-reviewer

  Verifies the plan covers every goal and authors acceptance criteria as per-task Test Expectations blocks. Reviews the plan artifact, not task implementations. Runs always (quick + full pipeline).

• qrspi-plan-test-coverage-reviewer

  Verifies that task test expectations cover all behaviors, edge cases, and error conditions, and that each expectation is specific enough to be verifiable. Reviews the plan artifact, not task implementations. Runs always (quick + full pipeline).

• qrspi-questions-reviewer

  Reviews questions.md for artifact quality only — no scope review (Questions has no scope-reviewer per canonical topology).

• qrspi-replan-analyzer

  Severity-classifies proposed changes from a completed phase. Reads fixes/, reviews/, tasks/ from disk and returns proposed-changes payload inline (orchestrator captures and passes to replan reviewer + scope-reviewer). Different role from qrspi-replan-reviewer.

• qrspi-replan-reviewer

  Reviews the replan-analyzer's proposed-changes payload for artifact-specific quality (correctness, clarity, completeness) per the QRSPI reviewer protocol. Scope/boundary review is handled by qrspi-replan-scope-reviewer.

• qrspi-replan-scope-reviewer

  Scope/boundary review for the replan-analyzer's proposed-changes payload. Reads skills/replan/owns-defers.md and applies the 3-check scope procedure. Companion to qrspi-replan-reviewer (which handles artifact quality).

• qrspi-research-collator

  Verbatim collation subagent — extracts the Summary block from each q*.md file and assembles them into research/_collated.md (staging file). The orchestrator renames _collated.md to summary.md. Mechanical extraction, NOT synthesis.

• qrspi-research-reviewer

  Reviews research/summary.md for artifact quality only — no scope review (Research has no scope-reviewer per canonical topology).

• qrspi-research-specialist

  Per-question parallel researcher. Answers an assigned research question with objective, factual findings and writes the report directly to disk. Research-isolation invariant binding — never receives goals.md or other-question content.

• qrspi-scope-tagger

  Derive one scope_tag per kept finding (post-verifier-fan-in) and emit reviews/{step}/round-NN-scope-set.txt. Multi-file artifacts: tag = referenced_files path. Single-file artifacts: tag = enclosing H2 heading text. Write-only — never mutates finding files or sidecars.

• qrspi-security-integration-reviewer

  Reviews merged code from multiple implementation tasks for cross-task security vulnerabilities. Individual task security was reviewed during Implement — this agent looks for issues that ONLY emerge when tasks are combined. Dispatched from the Integrate phase.

• qrspi-security-reviewer

  Identifies security vulnerabilities in the implementation. Runs after spec-reviewer passes, in parallel with other thoroughness reviewers.

• qrspi-silent-failure-hunter

  Identifies silent failures, inadequate error handling, and inappropriate fallback behavior. Runs after spec-reviewer passes, in parallel with other thoroughness reviewers.

• qrspi-spec-reviewer

  Verifies the implementer built exactly what the task spec requested — nothing more, nothing less. Used in both Implement phase (per-task code review) and Test phase (test code review). Gate reviewer — other reviewers only run if this passes.

• qrspi-structure-reviewer

  Reviews structure.md for artifact-specific quality (correctness, clarity, completeness) per the QRSPI reviewer protocol. Scope/boundary review is handled by qrspi-structure-scope-reviewer.

• qrspi-structure-scope-reviewer

  Scope/boundary review for structure.md. Reads skills/structure/owns-defers.md and applies the 3-check scope procedure. Companion to qrspi-structure-reviewer (which handles artifact quality).

• qrspi-test-coverage-reviewer

  Verifies tests are comprehensive and meaningful, covering behaviors, edge cases, and error conditions. Deep mode only. Runs after all correctness reviewers pass.

• qrspi-test-writer

  Dual-mode test-writing agent. In Implement-phase mode (task_definition present): writes per-task failing tests against the un-implemented spec. In Test-phase mode (task_definition absent): writes plan-level acceptance tests that verify the implementation meets the original goals. Does NOT modify pro

• qrspi-type-design-analyzer

  Analyzes type design for encapsulation, invariant expression, and correct use. Deep mode only, only when the task introduces new types. Runs after all correctness reviewers pass.

• qrspi-visual-fidelity-reviewer

  Per-task visual-fidelity reviewer for UI surfaces. Audits the implemented UI against wireframe reference artifacts and emits structured findings for material visual divergences. Wireframe-reference fidelity only — does not run screenshot diffing or audit non-UI surfaces.

Similar plugins