SKILL.md

Plugin SKILL.md from marketplace SKILL.md in BartSoj/SKILL.md. Commands read from the repo's manifest — 1 month ago.

Skills (38)

• ARCHITECTURE.md

  Design the system architecture — architectural style, component inventory, topology, cross-component flows, foundational ADRs, deployment shape, quality-attribute headlines, and evolution seams, at the structural level only. Use when asked to design the architecture, lay out components and flows, ma

• BEHAVIOR.md

  Model the system's behavioral contract — state machines per stateful aggregate, sagas for multi-step workflows with explicit compensations and timeouts, an idempotency-key registry, concurrency rules, temporal ordering invariants, and event emission timing under stable `SM-*` and `SAGA-*` IDs. Use w

• CLI_IA.md

  Generate a command-line information architecture registry — command inventory, grammar, global flags and environment, per-command blueprints, flows, exit-code taxonomy, terminal capability strategy, and use-case traceability. Use when asked to create a CLI IA, design the command tree, produce a comm

• CODE_REVIEW.md

  Review code changes using parallel subagents for security, bugs, quality, contracts, test coverage, and historical context. Use when asked to review code, review changes, do a code review, check code before merge, or produce a CODE_REVIEW.md.

• COMMAND_SPEC.md

  Produce a deep per-command design specification for a single command in a command-line application — argument-parsing rules beyond CLI_IA's grammar, stdin/stdout/stderr composition, every output mode's shape, table and block layout in human mode, interactive choreography (prompts, progress, pager) a

• DATA.md

  Design the persistent-data layer — datastore choice, ER diagram, schema per table/collection, indexes, access-pattern-to-index mapping, migrations, retention, sensitive-data inventory, consistency and concurrency strategy, and non-relational persistent assets. Use when asked to design the data model

• DECISION.md

  Resolve one or more Open Questions from any artifact by gathering evidence (web search, codebase grep, runtime sandbox, prior decisions), reasoning bias-free, and either deciding with an audit trail or handing off to the user via an asynchronous YAML mechanism — writing one `decisions/D-NNN-slug/DEC

• DESIGN_REVIEW.md

  Gate the design suite before continuous implementation begins — read DOMAIN, ARCHITECTURE, INTERFACES, DATA, BEHAVIOR, QUALITY, SECURITY, ERRORS end-to-end and catalogue every cross-artifact contradiction, every completeness gap that blocks downstream steps, every internal-quality defect, and every

• DOMAIN.md

  Model the domain — ubiquitous language, bounded contexts, entities, aggregates, value objects, domain events, services, and invariants with stable IDs. Use when asked to model the domain, create the domain model, define the glossary and entities, establish the ubiquitous language, or produce a DOMAI

• ERRORS.md

  Build the unified error / result taxonomy — classification tree, canonical error shape, alphabetical error-code registry with stable UPPER_SNAKE codes, wrapping rules across component boundaries, per-surface rendering, and a deprecation policy. Use when asked to create the error taxonomy, register e

• IMPLEMENTATION.md

  Execute an implementation plan task-by-task using test-driven development and delegate detail work to subagents. Produce an IMPLEMENTATION.md that records what was built, every deviation from the plan with rationale, and the final test-suite state. Use when asked to implement a plan, execute impleme

• INTENT_SPEC.md

  Produce a deep, per-intent design specification for one voice intent — full sample-utterance library, slot resolution detail, turn-by-turn dialog flow, confirmation flow detail, response composition with SSML hints, multimodal expansion detail, expanded failure matrix, context effects, and intent-sp

• INTERFACES.md

  Consolidate the machine-interface contract — global wire conventions, authentication, idempotency, versioning, endpoint catalogue with request/response shapes, event registry, and evolution policy under stable `EP-name` and `EVT-name` IDs. Use when asked to produce an INTERFACES.md, design machine i

• ISSUE.md

  File a defect-in-shipped-code trigger — a bug or regression — by reproducing the failure, comparing observed behavior against the source-of-truth artifact (INTERFACES, IA, SPEC), classifying severity, identifying root cause when known, and recording the audit trail in a complete `issues/<NNN>-<slug>

• MOBILE_IA.md

  Generate a native mobile information architecture registry — platform strategy, navigation model, deep-link strategy, permissions strategy, screen-template inventory, per-screen blueprints, flows, shared surfaces, push notifications, lifecycle behavior, and use-case traceability. Use when asked to c

• OPERATIONS.md

  Consolidate the system's operations contract — deployment topology per environment, CI/CD pipelines with promotion and rollback, environment strategy, config catalogue with stable `CFG_NAME` IDs, secrets inventory with rotation policies, integration catalogue with failure behaviour, runbooks for eve

• ORCHESTRATOR_CREATOR.md

  Design and create an orchestration agent that autonomously executes a multi-skill workflow. Use when asked to create an orchestrator, build a pipeline agent, design a workflow agent, automate a skill sequence, or produce an ORCHESTRATOR_CREATOR.md.

• PAGE_SPEC.md

  Produce a deep, per-page composition specification for a single web page — page shell, region layout, section composition, per-component layout detail, action inventory and placement, state-specific layouts, responsive behavior, page-specific motion, and accessibility notes — extending the lightweig

• PLAN.md

  Translate a specification into a sequenced implementation plan — ordered steps with inlined context (types, signatures, rules, constants, file paths), codebase citations for reuse and integration points, and per-step plus end-to-end verification. Use when asked to create a PLAN.md, plan implementati

• PROPOSAL.md

  Crystallize the product vision — problem, solution thesis, principles, non-goals, success criteria, target actors, scope boundaries, risks, and a glossary stub — as the first artifact in an Agent-Driven Development project. Use when asked to draft a proposal, write the proposal, start a new ADD proj

• PROTOTYPE.md

  Build a scratch prototype to empirically validate the `R-NN` risks surfaced by spec-review — write throwaway code under `units/<area>/u<NN>/prototype/`, run it, capture concrete evidence (output, dumps, timings, errors), and emit a verdict (`go`, `go-with-caveats`, or `abort`) that drives orchestrat

• QUALITY.md

  Design the system's observability and performance contract — logging standard, domain event log vocabulary, metrics catalogue with stable `METRIC-*` IDs, service-level objectives under `SLO-*` IDs with burn-rate alerts, tracing strategy, alert catalogue, performance budgets per flow, capacity envelo

• RECONCILIATION.md

  Reconcile a unit's SPEC and the top-level design suite with what implementation actually built — surface discrepancies, classify them, filter out implementation details, decide which discoveries are general truths about the system, and apply direct edits to SPEC.md, selected top-level artifacts, and

• ROADMAP.md

  File a planned-work trigger — a feature, refinement, refactor, or chore — by performing concept-overlap discovery against the design suite and existing roadmap, then synthesizing user intent into a complete `roadmap/<NNN>-<slug>/ROADMAP.md` with frontmatter, context, sketch, acceptance criteria, cur

• SCREEN_SPEC.md

  Produce a deep per-screen design specification extending one MOBILE_IA.md blueprint — screen shell, region layout, section composition, per-component layout detail, action inventory, gesture and interaction specifications, state-specific layouts, orientation and form-factor rules, platform variation

• SECURITY.md

  Model the system's threats and controls in one consolidated artifact — assets, hostile actors, trust boundaries, STRIDE threat catalogue, mitigations tied to implementation, authentication and authorization, secret management, input validation, abuse prevention, supply-chain security, incident respo

• SKILL_CREATOR.md

  Design and create a new skill, or refine an existing one, following project conventions. Use when asked to create a skill, design a new skill, write a skill definition, build a SKILL.md, scaffold a skill from requirements, improve an existing skill, or refine a skill's structure.

• SPEC_REVIEW.md

  Gate the unit SPEC before PLAN — read the trigger artifact(s) (roadmap or issue), the SPEC end-to-end, the unit's declared design read-set, every related unit's SPEC, and the codebase, then dispatch four parallel subagents (Scope & Risk Auditor, Compatibility Checker, OSS Library Scout, Internal Qua

• SPEC.md

  Produce an implementation-complete specification for a single work unit — file manifest, public interface contract with full signatures, ordered behavioral paths, error catalog bound to registered `ERR_CODE`s, test specifications, and design-layer citations (`INV-NN`, `EVT-name`, `SM-*`, `SAGA-*`, `

• SYSTEM_VERIFICATION.md

  Verify the composed application end-to-end by bootstrapping the full stack and running cross-cutting user scenarios. Use when asked to do system verification, test the full application, run end-to-end integration tests, verify the system works as a whole, or produce a SYSTEM_VERIFICATION.md.

• TRIAGE.md

  Analyze system verification failures, trace each to its originating design artifact, group related issues into fix batches, and produce an actionable fix plan. Use when asked to triage failures, diagnose integration bugs, analyze why the system broke, trace bugs to root causes, or produce a TRIAGE.m

• TUI_IA.md

  Generate a terminal UI information architecture registry — layout model, mode model, keybinding matrix, focus model, per-view blueprints, flows, shared surfaces, terminal capability strategy, and use-case traceability for rich full-screen terminal applications. Use when asked to create a TUI IA, des

• USE_CASES.md

  Enumerate every user-perspective capability of the system — actors, use cases with stable IDs, main and alternative flows, priorities, cross-cutting scenarios, and the surface mapping matrix — as the second artifact in an Agent-Driven Development project. Use when asked to create the use case catalo

• VERIFICATION.md

  Verify a running application by interacting with it as a QA tester. Use when asked to verify the app works, test the application, do QA testing, validate that implementation works, or produce a VERIFICATION.md.

• VIEW_SPEC.md

  Produce a deep per-view design specification for one view in a rich terminal UI application — view shell, per-panel layout, mode-specific layouts, keybinding detail with panel dispatch, focus mechanics, status-bar composition, state-specific layouts, terminal-capability degradation, mouse behavior,

• VOICE_IA.md

  Generate a voice information architecture registry — invocation model, slot model, dialog model and context, intent inventory, per-intent blueprints, multi-turn flows, shared surfaces, confirmation strategy, fallback and handoff, privacy and safety, platform strategy, and use-case traceability. Use

• WEB_IA.md

  Generate a web information architecture registry — page inventory, URL strategy, navigation, per-page blueprints, flows, and use-case traceability. Use when asked to create a web IA, design the site map, produce a page inventory, define URL strategy and navigation, or produce a WEB_IA.md.

• WORKFLOW.md

  Capture a conversational workflow into a structured, generalizable workflow document. Use when asked to capture the workflow, document what we just did, create a workflow from this conversation, summarize the workflow steps, or produce a WORKFLOW.md.

Subagents (1)

• add-orchestrator

  Autonomously execute the Agent-Driven Development (ADD) workflow — bootstrap a new project from a one-paragraph product idea through the full design suite, then drive continuous implementation by picking up roadmap items and issues. Use when asked to "run the full workflow", "build this project end

Similar plugins