kit

Plugin kit from marketplace dwarves-marketplace in dwarvesf/dwarves-kit. Commands read from the repo's manifest — yesterday.

Skills (16)

• kit-board

  Read the state of the boards and explain the daily brief, using the dwarves-kit `board` command. Triggers: "how are the boards", "how many boards are over threshold", "what is waiting in triage", "what needs my decision", "explain the brief", "is board sync healthy", "what did the bots do". Read-onl

• kit-precedent

  Look up what already exists before proposing a new tool, script, skill, cron, or memory note, using the dwarves-kit `precedent` command. Triggers: "should we build X", "do we already have something for X", "is there a script that does X", "have we done this before", or any plan that would add a new

• backlog-reconcile

  Use to audit a repo's `_meta/BACKLOG.md` Active queue against reality, "audit the backlog", "reconcile the board", "is the board still true", "dọn backlog", "rà soát backlog", or a scheduled backlog-reconcile run. NOT for filing new items onto the board, NOT for a single known-stale row (just fix it

• ci-drift

  Use for the whole-estate CI audit, "audit the CI", "are the runners clean", "check the release pipeline", "ci drift", "stale workflows / secrets / runners", or a scheduled CI-audit cadence run. NOT for one failing run (/kit:debug), NOT for doc prose (doc-drift).

• doc-drift

  Use for the whole-estate doc audit, "run the doc-drift loop", "audit the docs against the code", "are the docs still true", "doc drift sweep", "check every doc claim", or a scheduled doc-audit cadence run. NOT for diff-scoped doc sync inside a build cycle (/kit:docs), NOT for dated records like spec

• gauntlet-proof-audit

  Use to audit the gauntlet records, "are the ROUNDS.md records honest", "verify the proof corpus", "check each run record against its evidence", "audit the gauntlet corpus", "is the gauntlet proof trustworthy", or a scheduled run. NOT for re-running probes or rounds (`/kit:gauntlet` itself), NOT for

• get-api-docs

  Fetch curated API documentation using Context Hub (chub) before coding against any external API. Use when the task involves calling a third-party API (Stripe, OpenAI, Airwallex, Notion, etc.) to prevent hallucinated parameters and deprecated endpoints.

• loop-engineering

  Use to design or add a new bounded loop to the kit's SDLC orchestration ("design a new loop", "is this worth a loop"), or mentions the Karpathy loop / autoresearch / hill-climb / search-and-select prompt or parameter tuning. NOT for a one-off in-session Stop-hook goal (use goal-craft), NOT for the d

• memory-tidy

  Use when auditing, consolidating, or cleaning a repo's .claude/memory store, "dọn memory", "memory tidy", the biweekly memory audit, duplicate or stale memory notes, MEMORY.md index drift, or a scheduled memory-tidy audit. NOT for the built-in machine-local auto-memory under ~/.claude/projects, NOT

• observe

  Query and render the kit's control plane (fleet runs, spend, debt; backed by lib/bench) from an agent session. NOT for your own Claude Code usage, skill, tool or hook stats (that is `session observe`, lib/session/observe). Use when asked about the fleet's runs, gate verdicts, conformance, spend/toke

• repo-hygiene
• skill-review
• stats
• system-design
• topology-drift
• web-drift

Slash commands (39)

• /kit:absorb

  Maintainer-only: audit the kit's upstream sources (Credits drift + seed-rescan) and draft a dated, proposal-only absorption report. Does not execute, does not merge.

• /kit:adopt

  Adopt the current (or a target) repo into the dwarves-kit operate-contract: write a small AGENTS.md pointer (never over a repo's own file) + a CLAUDE.md loader + a WORKFLOW pointer + the proof marker, idempotently, and wire the lane/loop-type/proof classifiers so the ship-gate engages.

• /kit:assign

  Turn a backlog item (ID-NNN) into a scoped goal draft and route it into the right WORKFLOW lane. Writes .claude/goals/, never executes.

• /kit:battery

  The full independent-verification battery for a finished branch: a fresh-context acceptance verifier that RE-EXECUTES the verification commands against a stated baseline, a multi-lens review, and the kit:advisor extra lens, dispatched in parallel at prescribed model tiers, findings merged into one v

• /kit:debug

  Systematic debug loop: root cause before any fix. Four phases, an evidence ledger, the 3-fix architecture wall. Use for any bug, defect, regression, test failure, or unexpected behavior before proposing a fix: 'fix this bug', 'why does X fail', 'the test is failing', 'it broke after Y', 'debug X', '

• /kit:design

  Opt-in interactive solution-design beat between /think and /spec. Use when the operator wants to shape HOW to build after the idea passed /kit:think: 'design the solution', 'how should we build X', 'compare approaches', 'architecture for X', 'thiết kế giải pháp', 'thiết kế kiến trúc', 'so sánh cách

• /kit:devs-team

  Parallel multi-lens critique of a solution design (the active spec if present, else the decision brief). Dispatches 5 engineering lenses, merges findings, reports a verdict. Report-only, never blocks.

• /kit:dispatch

  Fire several disjoint VALIDATED specs concurrently, each in its own worktree, then converge. Cross-goal fan-out behind a disjointness gate + drift guard; lead-owned merge, no DAG.

• /kit:docs

  Update all project documentation to match the current codebase. Cross-references the diff against every doc file and fixes drift.

• /kit:draft-agent

  Meta-agent agent-builder. From a one-line description, generates a new subagent definition OR a mega-goal sub-goal file and (by default) installs the subagent so it is dispatchable next session. --draft stops at a staged draft for review.

• /kit:execute

  Whole-spec execution with verification. Dispatches one builder for the whole spec, verifies every task's criteria once at the end, retries fixable failures (max 2), names what did not ship.

• /kit:explain

  Turn a merged change into a literate-diff explainer a human READS to understand: background -> goal + intuition -> a prose-ordered diff -> a diagram. Grounds the material mechanically, then hands it to the operator's configured teacher seam; grounded in the actual diff + test results, never the agen

• /kit:feature-map

  Build or refresh a source-cited, agent-checkable feature inventory for ANY target project you point it at: per-module spec + a top-level checklist. Works standalone (what does this codebase do) or as a migration source of truth (what needs porting). Refresh mode is an audit-loop instance, sibling to

• /kit:gauntlet

  Probe-convergence engine: a bounded-revise loop that converges an ARTIFACT (docs, a runbook, a spec, an API surface) toward a FIXED OUTCOME by making a fresh clean-room probe agent attempt the outcome contract unaided each round; when the probe fails, the artifact gets revised, the environment torn

• /kit:greenlight

  Post-push CI-green lane: snapshots an open PR's checks via gh, classifies each failure real vs flaky, fixes real ones via the kit:fix-agent shape (verified locally before push), retries flaky checks within a bounded budget, and reports exactly one terminal state. Opt-in, report-only -- it never hard

• /kit:grill

  Universal intake interview: one type-shaped question at a time, each with a recommended answer, until the task is actually understood. Answers are written where they live (glossary, sparse ADRs, the goal draft's Context) the moment they resolve. Runs between type classification and the phase-0 Done=

• /kit:kit-health

  Run a self-assessment of the kit against its own philosophy. Checks file count, hook performance, source citations, and structural health.

• /kit:mega

  Turn a multi-objective destination into a sequenced roadmap of dependent sub-goals: decompose, front-load every clarification once, set the per-run merge config, then hand off to the bounded /goal loop. Ship-layer auto-merge rides the ship-gate; it never bypasses it.

• /kit:next

  Pick up the next undone task from the spec. Loads context, shows acceptance criteria, lets you drive the implementation.

• /kit:onboard

  Guided first-run: detect the install mode, offer /kit:adopt for this repo, pick modules, capture the consumer knobs that make them work, disclose the plugin-path gaps honestly, and end with a two-idea tour (lane, proof of done) plus a menu. Previews and confirms every write; a decline is a no-op.

• /kit:pitch

  Assemble an outward buy-in doc from what a gated run already produced: the spec, the proof-of-done, the implementation-notes, and the gate ledger's grill/DEBT records. Outcome-first, 5 sections, ending in an ask. Never generates new analysis; a missing source is reported honestly, never invented.

• /kit:prototype

  Opt-in throwaway-spike beat beside /kit:design. Builds throwaway code that answers ONE design question: a logic/state model driven by hand in a TUI, or 3-5 structurally different UI variants on one route. The decision folds into the brief/spec; the prototype survives on a prototype/<name> branch, ne

• /kit:quiz-gate

  The ★-tap NUDGE before merging a significant+worthy gate PR: a 5-question quiz built from the ACTUAL diff+tests, handed to the operator's configured teacher seam. Three logged responses (engage/defer/accept); advisory, never must-pass. Gates the human's attention, not the merge.

• /kit:retro

  Run a retrospective after shipping. Captures what worked, what didn't, and action items for the next cycle. Outputs to docs/retro/RETRO.md.

• /kit:review-team

  Parallel code review with 3 specialist lenses plus the kit-default kit:advisor extra lens. Dispatches the security, architecture, and test-coverage lenses simultaneously, adds the cross-cutting kit:advisor (critique mode), then merges findings.

• /kit:review

  Paranoid code review. Security, architecture, regressions, missing tests, edge cases. Produces actionable TODOS.

• /kit:ship

  Ship: review gate, tests, version bump, changelog, conventional commit, docs update, PR. Complete pipeline from done to merged.

• /kit:spec-validate

  Adversarial review of a spec before implementation. 7 specialist lenses attack the spec from different angles (6 advisory, 1 blocking on the design record).

• /kit:spec

  Generate a development spec from a feature idea or decision brief. Creates docs/specs/ with structured requirements.

• /kit:start

  Detect project state and suggest the right next command. The entry point for any session.

• /kit:test-plan-review-team

  Parallel multi-lens adversarial critique of a spec's test plan (the ## Test plan section), with a bounded revise loop that tightens it. Dispatches 6 test-design lenses, merges findings, optionally revises, reports a verdict. Report-only, never blocks /kit:execute.

• /kit:test-plan

  Derive a test-case coverage matrix from a spec's acceptance criteria before /kit:execute. Writes a `## Test plan` section into the active spec. A coverage target across categories, not an exhaustive test list.

• /kit:test-write

  Turn a SOLID-verdict `## Test plan critique` into real, executing test code via `kit:test-writer`, one case per matrix row. Never dispatches against a missing, stale, or non-SOLID verdict.

• /kit:think

  Challenge an idea before writing any spec. 6 forcing questions that reframe the product. Use when the operator brings a new idea, feature, or product to stress-test: 'think about X', 'should we build X', 'is X worth it', 'brainstorm X', 'design X' (this runs first, /kit:design follows), 'thiết kế X'

• /kit:ui-design

  Downstream UI-design loop: write a structured ## UI design brief, delegate generation to the external frontend-design skill, critique via /kit:visual-team, and auto-revise (bounded). Opt-in, downstream-facing; the kit ships no renderer.

• /kit:verify

  Re-run the test levels (kit:task-verifier + kit:integration-verifier + kit:acceptance-verifier + kit:system-verifier) on the current spec/branch read-only, no rebuild. The on-demand executor of the complete V-model right arm.

• /kit:visual-team

  Parallel multi-lens critique of a visual/UI design. Dispatches 5 design lenses, merges findings, reports a verdict. Report-only, downstream-facing (the kit itself has no UI).

• /kit:wayfind

  Plan a chunk of work too big for one agent session as a shared decision map: map.md + typed decision tickets in the mega-goal folder, resolved one per session through the kit's own grill/prototype/research machinery, until nothing is left to decide. Then hand off to /kit:spec or a ROADMAP, never str

• /kit:wrap

  The session-scoped landing step after ship: flips board rows, merges green PRs, checks deploys, tidies worktrees, prints the skim-first report. Use when the operator says to wrap up, close out the session, or land the work: 'wrap up', 'close out', 'session wrap', 'land it', 'tổng kết session', 'wrap

Subagents (31)

• acceptance-verifier

  Dynamically executes the active spec's acceptance criteria via its `## Verification` section and reports whether the build actually satisfies them. Fills the agent-less Acceptance row of the V-model right arm. Read-only -- cannot modify the codebase.

• advisor

  The single cross-cutting generic review lens. Runs in TWO modes at the final integration/UAT boundary -- critique (an extra uniform lens ON TOP of the specialized per-phase reviewers) and over-suggest (proposes additional ideas/sub-goals to improve the work, surfaced to the human just before the fin

• agent-effectiveness

  Validates an agent definition's EFFECTIVENESS (not just its structure) across four lenses -- tools minimal-yet-sufficient, description triggers right, instructions produce a good result, model tier fits. Dispatched diff-keyed on new/changed agent defs at the agent-author phase. Read-only, advisory,

• api-reviewer

  Reviews a diff through the API-CONTRACT lens only (breaking changes, versioning, request/response schema, error codes, backward compat, idempotency). Read-only. Dispatched by /kit:review-team as the api domain lens when the diff touches a public interface.

• audit-scanner

  Shared read-only Tier-2 evidence scanner for audit-loop instances (doc-drift, topology-drift, ci-drift, backlog-reconcile, web-drift, repo-hygiene, gauntlet-proof-audit, future ones). Dispatched by an audit skill with a target set + a contract + an evidence-class instruction; returns per-item findin

• break-it

  The adversarial prober lens. Given a branch whose behavioral code carries a green test suite, it hunts for a concrete INPUT or CALL SEQUENCE the suite does not constrain -- one that changes behavior without failing any test. Escalation-tier lens dispatched from /kit:battery when the diff carries beh

• brief-reviewer

  Statically reviews a design brief or requirement (DECISION-BRIEF.md, a spec's Problem/Context section, or an equivalent requirement doc) for clarity, completeness, and testability before it hardens into a spec. Read-only -- cannot edit the brief.

• claim-verifier

  Adversarially verifies an ARBITRARY free-text claim before it is trusted. Runs an in-harness panel of N independent skeptics, each told to REFUTE the claim (default-refute-if-uncertain, fail-closed), then returns a STRUCTURED majority-vote verdict (HOLDS / REFUTED, how many refuted, the threshold, p

• code-reviewer

  Focused code reviewer. Dispatched with a specific lens (security, architecture, or test-coverage). Read-only. Used by /review-team for parallel review.

• data-etl-worker

  Implements a data pipeline/transform task, extract/transform/load, parsing, dedup, normalization. Write-capable; prefers DuckDB SQL for the transform per the house stack. Dispatched by /kit:execute as the builder when `role-classify.sh agent-for` names it for the data-etl domain.

• db-migration-worker

  Implements a schema migration task, the up migration plus its reverse/rollback, backfill, and index changes. Write-capable. Dispatched by /kit:execute as the builder when `role-classify.sh agent-for` names it for the db-migration domain.

• devops-triage

  Triages a production error alert (service name, error sample, optional deploy sha) into a bounded root-cause verdict, gathering evidence via Cloudflare Workers Logs history and git log/diff/show around the deploy sha. Dispatched on-demand for "triage this production alert/error", "why is <service> e

• doc-verifier

  Verifies that documentation matches the code. Dispatched by /docs after it updates docs, before the commit. Read-only -- cannot edit docs or code. Checks doc claims against the live codebase, the Docs-phase twin of kit:task-verifier.

• fix-agent

  Applies targeted fixes based on feedback from kit:task-verifier or the other end verifiers (integration, acceptance). Scoped to specific files and specific issues. Does not add features or refactor.

• frontend-reviewer

  Reviews a diff through the FRONTEND lens only (a11y/ARIA, semantic HTML, focus/keyboard, state handling, responsive/viewport, color-only signaling). Read-only. Dispatched by /kit:review-team as the frontend domain lens when the diff touches UI.

• infra-reviewer

  Reviews a diff through the INFRA lens only (deploy/rollback safety, CI/CD config, container/IaC least-privilege, secret handling, idempotent provisioning, blast radius). Read-only. Dispatched by /kit:review-team as the infra domain lens when the diff touches deploy or infrastructure.

• integration-verifier

  Verifies that the tasks of a completed build actually wire together. Dispatched once at /execute Step 4 for multi-task specs. Read-only -- cannot modify the codebase. Checks cross-task wiring + global acceptance, not per-task acceptance.

• meta-agent

  The agent that drafts agents. From a one-line description, drafts a new subagent definition OR a new mega-goal sub-goal file, matching the kit's exact frontmatter + structure. Output is always a DRAFT for human review; it never self-installs or self-runs.

• performance-reviewer

  Reviews a diff through the PERFORMANCE lens only (hot paths, N+1, allocations, caching, latency, complexity). Read-only. Dispatched by /kit:review-team as the performance domain lens when the diff touches performance-sensitive code.

• recheck-verifier

  Fresh-context re-audit of a right-arm verifier's PASS (kit:task-verifier / kit:integration-verifier / kit:acceptance-verifier / kit:system-verifier). RE-EXECUTES the recorded verification command in a fresh context and re-judges the outcome; it is NOT a read-back of the recorded run-table. Read-only

• research-architecture

  Maps architecture patterns and conventions in an existing codebase. Dispatched by /spec for brownfield projects. Read-only.

• research-context

  Quick brownfield orientation for a target area (endpoints, data models, UI, test coverage, recent history) -- capped at 80 lines, throwaway context for a spec about to be written. Dispatched by /spec, /test-plan. Read-only.

• research-features

  Deep, uncapped, source-cited feature inventory for one module of a target project -- every entry point gets a file:line citation and a behavior contract; adds a MIGRATE table + parity contract when the request names a port/migration target. The definitive deliverable, not throwaway context (see kit:

• research-pitfalls

  Finds landmines and risks in a codebase area before new work begins. Dispatched by /spec for brownfield projects. Read-only.

• research-stack

  Maps the technology stack of an existing codebase. Dispatched by /spec for brownfield projects. Read-only.

• responding-to-review

  Responds to code review feedback with technical rigor, not performative agreement. Verifies before implementing. Pushes back when reviewer is wrong. Read-only by design (proposes fixes; the user decides whether to apply).

• security-reviewer

  Deep security review of code changes. Read-only. Dispatched by /review-team for focused security analysis. More thorough than /review's built-in security checks.

• slop-stripper

  Behavior-preserving AI-slop strip pass. Given a base ref, scans the branch diff and applies surgical edits that strip AI-generated smell (redundant comments, over-defensive handling, unnecessary casts, flattenable nesting, patterns inconsistent with the file) before merge. Never changes behavior unl

• system-verifier

  Runs the whole assembled project's test suite end to end as the dynamic right-arm mirror of the design phase. Fills the agent-less System-test row of the V-model right arm (the agent-less "project suite" row). Read-only -- cannot modify the codebase.

• task-verifier

  Verifies a completed task against its spec acceptance criteria. Callers invoke it per task or, under /kit:execute, once over every task of the build. Read-only -- cannot modify the codebase.

• test-writer

  Turns a reviewed test-plan coverage matrix into runnable test code, one case per matrix row, in the repo's existing test framework. Write-capable. Dispatched by /kit:test-write after a SOLID `## Test plan critique` verdict, before the real test run becomes the verifier.

Hooks (28)

• SessionStart

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/post-compact-reinject.sh"

• SessionStart

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/context-readiness.sh"

• SessionStart

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/codebase-index.sh"

• SessionStart

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/backlog-stage.sh" --surface

• SessionStart

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/kit-update-notice.sh"

• PreToolUse

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/safety-gate.sh"

• PreToolUse

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/ship-gate.sh"

• PreToolUse

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/spec-drift-guard.sh"

• PreToolUse

  "${CLAUDE_PLUGIN_ROOT}/hooks/secrets-guard.sh"

• PreToolUse

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/commit-format.sh"

• PreToolUse

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/board-row-gate.sh"

• PreToolUse

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/batch-debt-warn.sh"

• PreToolUse

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/money-gate.sh"

• PreToolUse

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/tool-policy-guard.sh"

• PostToolUse

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/auto-format.sh"

• PostToolUse

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/output-offload.sh"

• PreCompact

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/pre-compact-backup.sh"

• PreCompact

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/harvest.sh"

• Stop

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/anti-rationalization.sh"

• Stop

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/slop-cleaner.sh"

• Stop

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/citation-guard.sh"

• SessionEnd

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/backlog-stage.sh"

• SessionEnd

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/harvest.sh" --lab-log

• UserPromptSubmit

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/context-hints.sh"

• UserPromptSubmit

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/prose-rag.sh"

• UserPromptSubmit

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/context-budget.sh"

• Notification

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/notification.sh"

• PermissionRequest

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/anchor-root.sh" "${CLAUDE_PLUGIN_ROOT}/hooks/permission-auto-approve.sh"

Similar plugins