ainb-toolkit

Autonomous product factory — /make-a-goal in godmode. Point it at a north-star outcome and it runs the whole delivery machine unattended: brainstorm the feature landscape (Discover), put every idea through a Feasibility Court, cluster into a Roadmap of epics, then per epic Plan → adversarial review → Execute (model-paired build) → VERIFY (drive the real UI/TUI/API, not mocks) → ship a stacked PR, looping until the backlog is dry or a budget/time bound fires. One human gate (roadmap blessing), hard stop rules, always-on tabbed RAG dashboard, crash-proof resume from state files. Use when the user says 'godmode', 'run the factory', 'ship X to nirvana', 'execute this whole backlog autonomously', 'goals within goals', or wants a multi-epic programme driven end-to-end with real end-to-end verification. NOT for single features, one-off verification, or scoped builds — use /make-a-goal, browser/tmux verification skills, or a plain plan for those.

Sync user-level agent config changes back to the ainb-toolkit repo (works for Claude, Codex, Copilot)

Produce a self-contained, richly styled HTML explainer for any topic the user asks about. Picks the right template from a bundled set of 22 visual patterns (feature explainer, concept explainer, module map, PR review, ADR, options paper / trade-off analysis, system diagram, flow- chart, status report, slide deck, prototype, editor, etc.), fills it with real content, augments with inline diagrams via sister skills (/fireworks-tech-graph for architecture / flow / sequence diagrams, /graphify for knowledge graphs), applies a Claude-brand polish layer, and publishes it: by default onto the configured here.now custom domain (path mount + searchable categorised index + password lock per the config's protect rule, driven by ~/.herenow/explainers.json), or a plain here.now URL, or a GitHub gist (--gist / --gist --public). Local-only output is available with --local. Use when Stevie says "/explain-to-me", "explain-to-me X", "make me an explainer for X", "give me an HTML explainer", "render this as a webpage", "ADR for

Self-improvement through conversation analysis. Extracts learnings from corrections and success patterns, permanently encoding them into agent definitions. Philosophy - Correct once, never again.

Claude Code observability skill: analyze session traces stored in Langfuse, extract learnings from corrections, identify success patterns, and propose agent/skill improvements based on historical data. Powers self-improvement through trace analysis of Claude Code sessions.

Create well-formatted git commits for changes made during the session

Generate a handover document for transferring work to another developer or spawning an async agent

Interview the user about a plan file to extract detailed requirements, clarify ambiguities, and uncover edge cases. Uses iterative questioning to produce a comprehensive specification.

Conduct comprehensive multi-source research (codebase, docs, web) by spawning parallel sub-agents and synthesizing a single cited document. Searches past learnings first, then live codebase, docs, and optionally web. Use for "research X", "how does X work in this repo", "investigate", "find prior art", "compare approaches", "gather sources on".

Read-only branch-scoped situation report for Stevie. Surfaces beads in-flight, what's been done since last invocation, what's pending, active swarms/coding-agents/subagents in this worktree, and — most importantly — what specifically needs Stevie's input (decisions, PR merges, AskUserQuestion threads, blocked beads). Output is progressive tables, one shape per stage, address the user as Stevie. Trigger on /standup, "give me a standup", "what's the state of this branch", "what do you need from me". Read-only — never claims beads, never merges, never posts anywhere.

Verify implementation against specifications

Turn an idea into an approved design spec. Brainstorm is the orchestrator - it scaffolds a topic-stub with subject-type detection and ASCII preview hypotheses, then DELEGATES the iterative Q&A to /interview. After /interview returns a spec, brainstorm self-reviews, gates on user approval, and hands off to /plan. No HTML, no browser - terminal + ASCII previews only.

Get a Distinguished Engineer level technical critique of the current approach

Execute an approved implementation plan from plans/ step-by-step — phase-by-phase code changes, run success criteria, check off boxes, handle checkpoints and parallel waves

Produce a fully-filled autonomous-run mega-prompt at .agents/goals/<slug>.md by interviewing the user for the missing context. The output file is the "set the goal · walk away · come back to shipped work" prompt — drop it into a fresh Claude Code / Codex session to run end-to-end without hand- holding. Use when Stevie says "/make-a-goal", "/make-a-goal <outcome>", "make a goal", "set a goal", "build me a goal prompt", or similar. The skill does NOT execute the goal — it only produces the artifact. Named `make-a-goal` (not `goal`) because `/goal` is a reserved system slash.

Create a detailed implementation plan

Create a test-driven development plan for the project

Manage and search the research cache for previously analyzed repositories

Intelligence-first model routing for LLM pipelines. Sonnet is the default. Haiku is the exception, requiring positive justification. Opus for high-stakes and complex reasoning. Protects answer quality while reducing cost 30-50% vs defaulting everything to the session model. Use when: building multi-model pipelines, optimising API costs without sacrificing quality, designing retry strategies, implementing prompt caching, choosing models for sub-tasks.

Create or update AgentSkills. Use when designing, structuring, or packaging skills with scripts, references, and assets.

Analyze Stitch projects and synthesize a semantic design system into DESIGN.md files

Delegate coding tasks to subagents or run Claude Code/Codex in tmux sessions. Use the Task tool for focused multi-step coding work. Use tmux for long-running interactive sessions, parallel worktree-based fixes, and PR reviews. NOT for simple single-file edits — do those directly.

Manage and track installed plugins, skills, and extensions across all sources

Enterprise SRE patterns for AI agent operations. Provides cost caps, circuit breakers, stall detection, observability, and runbook-driven incident response for autonomous agent workflows. Use when: (1) Running long autonomous agent sessions, (2) Managing multi-agent swarms, (3) Monitoring agent costs and performance, (4) Debugging stuck or expensive agent loops, (5) Setting up agent observability.

Spin up disposable email inboxes via myagentinbox.com for tests, signup flows, OTP/magic-link captures, and any task that needs to receive mail from a service without using a real mailbox. Provides a REST-first bash workflow (no MCP required), 24h-lifetime inboxes, polling helpers with timeout, and verification-code/magic-link extractors. Use when the user mentions "throwaway email", "disposable inbox", "test signup", "OTP capture", "magic link", "verify email flow", "/expect-test signup", or any flow where a service emails a code and the agent needs to read it back. Stores inbox metadata in a tool-neutral state file under .agents/agentmail/ so multiple inboxes survive across sessions. MCP is intentionally NOT recommended — direct REST is simpler and avoids the npx mcp-remote shim. mcporter is documented as a fallback for projects already standardised on MCP tooling.

Professional ATS (Applicant Tracking System) resume matching and scoring tool that operates with the precision of enterprise systems like Greenhouse, Lever, Workday, and Breezy HR. Use this skill when: (1) Matching a resume against a job description to calculate fit scores (2) Analyzing resume-JD alignment with detailed category breakdowns (3) Identifying gaps between candidate qualifications and job requirements (4) Getting actionable suggestions to improve resume match percentage (5) Preparing a resume for ATS optimization before job applications Supports PDF, DOCX, Markdown, and plain text inputs for both resumes and job descriptions.

Attach to Agent Session

Six proven autonomous agent loop patterns with guard rails. Provides reusable patterns for generate->validate->fix, explore->hypothesize->test, and other autonomous workflows. Includes the reviewer-never-authored principle for quality assurance. Use when: (1) Building autonomous agent workflows, (2) Designing self-correcting pipelines, (3) Implementing agent retry/fix loops, (4) Setting up multi-agent review processes, (5) User asks about agent loop patterns.

Combined AI browser testing + visual inspection. Runs expect-cli for automated test generation/execution, then debug-bridge for screenshots and DOM inspection. Use when you need to verify UI changes with both automated tests AND visual evidence. Trigger on "verify in browser", "browser verify", "visual test", "test and screenshot", or after completing UI fixes that need proof.

Remove Agent Worktree

Comprehensive cryptocurrency market research and analysis using specialized AI agents. Analyzes market data, price trends, news sentiment, technical indicators, macro correlations, and investment opportunities. Use when researching cryptocurrencies, analyzing crypto markets, evaluating digital assets, or investigating blockchain projects like Bitcoin, Ethereum, Solana, etc.

Browser automation and inspection for AI agents via WebSocket

Discuss and explore ideas before committing to an approach

Work on GitHub issues systematically with proper development workflow

AI-powered browser testing using expect-cli. Auto-detects dev server, reads git diff, generates and executes browser tests via Playwright. Use when you need to verify UI changes in a real browser, test user flows, or validate fixes visually. Trigger on "test this in the browser", "verify the UI", "run expect", "browser test", or after completing UI fixes.

Expose a local service on Tailscale with a unique path

Analyze codebase and identify missing test cases, then create GitHub issues for each gap

Generate zero-dependency HTML presentations with keyboard navigation, print CSS, responsive design, and speaker notes. Creates a single self-contained HTML file — no build step, no framework, no CDN. Use when: (1) Creating slide decks or presentations, (2) Building pitch decks, (3) Making technical talks, (4) User asks for slides or a presentation, (5) Quick visual content for meetings.

Analyze and fix the specified GitHub issue

Safely split, reorder, or rewrite already-pushed commits using a throwaway worktree. Isolates the surgery from any existing checkout, verifies byte-identical tree before force-pushing, and uses --force-with-lease to refuse silent clobbers from concurrent work. Use when asked to split a pushed commit into atomic pieces, fix a bulked commit after the fact, or reshape recent history on a shared branch.