a5c-ai/babysitter

Agent Dispatch builds dependency-aware batched queues for parallel agent execution. Use it to order feature work by DAG, respect concurrency limits, and assign agent types safely.

Architecture Design defines a system’s components, technology stack, data model, APIs, and deployment considerations. Use it when turning product requirements into an implementation-ready architecture and ADRs.

Provides system and API design guidance covering component boundaries, data flow, integration patterns, and scalability. Useful when making architectural decisions in agentic AI workflows.

Searches public repositories for SKILL.md files, classifies projects by archetype, and extracts transferable processes, plugin ideas, and procedural insights for the babysitter process library. Use when exploring external workflows, discovering new skills, or surveying the skill landscape.

Atlas turns a stated need into a real systems atlas by scanning Azure, git repos, and local directories, then enriching the findings with an Atlas knowledge graph. Use it to inventory systems, mine processes or data, and collect real constraints or gotchas.

Atlas Graph Query provides reference commands for searching and inspecting the Atlas knowledge graph through MCP tools. Use it after scanning real sources when you need comparison context, node lookups, neighbors, kinds, clusters, stats, or wiki pages.

Babysit orchestrates structured tasks through the Babysitter SDK and run lifecycle. Use it when a task needs explicit process steps, loops, or an orchestrated run.

Fetches open issues assigned to a5c-agent from the babysitter repository and orchestrates their resolution by delegating each to an AI implementation skill, creating a branch and pull request for each fix. Use when you want to automatically process and close all assigned issues.

Orchestrates iterative runs using the Babysitter SDK, enabling automated process management and workflow execution for AI agents. Supports interactive and non-interactive modes.

Discovers public GitHub repositories that import from @a5c-ai/babysitter-sdk and maintains an additive catalog of adopters. Use to find external usage examples or monitor adoption.

Scaffold complete Python CLI applications using Click with proper project structure, command groups, decorators, and Rich terminal output.

Code Generation writes the least code needed to satisfy requirements while matching existing project patterns. Use it during TDD green and refactor phases or when applying fixes in a debug workflow.

Code Review evaluates correctness, security, performance, maintainability, and test coverage in implementation results. Use it to get an overall score, findings, technical debt, and action items.

Review supplied code changes for correctness, security, tests, performance, and style, then score them against a configurable quality threshold.

Code Review Patterns assesses code for security, quality, performance, and maintainability, with confidence-gated findings and a required router contract. Use it during build, debug, or full review workflows.

Integrates crisis response platforms for real-time monitoring, rapid detection, coordinated response, and comprehensive documentation.

Debugging Patterns provides log-first investigation, git bisect correlation, and pattern-based diagnosis with confidence scoring. Use it during debugging, unexpected build failures, and error-handling reviews.

Feature Intake parses feature descriptions, images, and screenshots into structured requirements. Use it to extract requirements, classify feature type, estimate complexity, and draft acceptance criteria.

Automatically identify and fix failing CI/CD workflows on the staging branch using the babysitter orchestration skill. Ideal for developers who want to quickly resolve pipeline failures without manual intervention.

Integrate Harness adds a production-quality adapter for a new CLI-based coding agent. Use it when adding adapter scaffolding, session parsing, auth detection, hooks or plugins wiring, tests, and docs.

Kanban Management initializes and tracks a Kanban board for feature workflow planning. Use it to place features into columns, set WIP limits, assign priorities, and snapshot board state.

PRD Creation turns interviews and product briefs into a structured product requirements document. Use it to define goals, features, acceptance criteria, MVP scope, and prioritization.

Planning Patterns provides a structured method for researching, brainstorming, and turning requests into phased plans. Use it before execution when you need alternatives, risks, acceptance criteria, and a handoff plan for building.

Scaffold new babysitter process definitions following SDK patterns and best practices. Guides a 3-phase workflow from research to implementation.

Product Brief Creation turns market, domain, and technical research into a structured product brief. Use it to define the problem, target users, value proposition, success metrics, and scope boundaries.

Project Documentation generates comprehensive project docs from a source tree, including component inventories, onboarding context, deep-dive pages, and Mermaid architecture diagrams. Use it when you need to document an existing codebase for new contributors or maintainers.

QA Testing generates and runs API and end-to-end tests, then reports coverage and a quality gate result. Use it after implementation or code review to verify behavior and catch regressions.

Move a small project from a quick technical specification through story-by-story implementation, tests, and scored code review with BMAD Quick Flow.

Analyzes a babysitter run from an external repository, producing a retrospective with insights on process quality, failure patterns, and improvement suggestions. Use for reviewing how others use babysitter or debugging public runs.

Plan a BMAD implementation sprint by choosing ready stories, setting a coherent goal, ordering dependencies, and identifying blockers.

Story Development implements user stories using test-driven development. It reads acceptance criteria, writes failing tests, implements the minimum code, and verifies the result with the full test suite.

Streaming Progress emits formatted progress events with timestamps, stage data, and completion percentages for UI streaming. Use it when an execution stage needs machine-readable updates and a human-readable summary.

Test Automation runs Vitest and Playwright suites, collects pass/fail counts, coverage, and failure details. Use it when you need automated test execution and root-cause analysis for changed code.

Implements UVM-based hardware verification testbenches including stimulus generation, scoreboarding, coverage-driven verification, and assertion-based testing.

UX Design creates user flows, wireframes, and a basic design system from a PRD. Use it when planning information architecture, interaction patterns, accessibility, and developer handoff specs.

Worktree Isolation manages git worktree creation, isolation checks, and cleanup for parallel development. Use it when you need separate branch workspaces for feature work, testing, or merges.