talentedgeai
OrganizationUniversal 8-agent templates for Infinite Leverage bootstrap skills
Categories
Indexed Skills (61)
developer-daily
Weekdays at 9:03 AM — reads project-status.html for today's approved items, invokes dev-planning to build the day's task list, then runs the development loop (dev-tdd per task) until blocked or all tasks complete. Opens a PR for each completed task.
devops-daily
Weekdays at 6:03 AM — checks production health (Vercel deployment status, uptime), reviews GitHub Actions CI/CD runs for failures, surfaces open PRs with failing checks, documents any issues into epic-status.md before the PM reads it at 7 AM.
email-marketer-weekly
Thursdays at 10:03 AM — checks the email index, drafts this week's newsletter based on the most recently published post, saves to emails/drafts/newsletter-YYYY-MM-DD.md. Draft only — does not send without approval.
pm-daily-plan
Weekdays at 7:03 AM — reads epic-status.md to extract the current to-do list, checks git log and standup check-ins, sets today's 3 priorities, auto-approves low-risk tasks, writes updated project-status.html for the developer agent to pick up at 9 AM.
pm-eod-summary
Weekdays at 6:37 PM — summarises what shipped today, what is blocked, and what is queued for tomorrow. Updates docs/project-status.html. Notifies team via Lark if configured.
pm-standup-compile
Weekdays at 6:07 PM — reads all standup check-ins, compiles daily team briefing to standup/briefings/YYYY-MM/YYYY-MM-DD.md, notifies team via Lark if configured.
pm-weekly-rag
Fridays at 5:07 PM — generates Red/Amber/Green status report from the week's briefings, writes to docs/product/rag-YYYY-MM-DD.md, notifies team via Lark if configured.
web-publisher-weekly
Wednesdays at 9:03 AM — finds content/topics/ folders that have blog.md and a hero image but no committed .jsx page, builds the Next.js page, optimises the image to WebP, updates the blog index, commits. Does not push — operator pushes manually.
dev-brainstorm
Opens up the options before committing to an approach — explores different ways to solve a problem, compares trade-offs, and helps choose the best path forward. Use when you're unsure how to build something or when there are multiple reasonable solutions.
dev-handoff
Writes a structured summary document (HANDOFF.md) so work can be picked up again without losing context — covering what was done, what's still in progress, what's blocked, and what the next person needs to know. Use whenever a working session ends before a feature is complete.
dev-improve-arch
Refactors a messy or over-complicated section of the codebase with a clear before/after plan. Use when tech debt is slowing down development or when a module has become hard to understand or change. Scans the problem area, ranks the improvements, and requires explicit approval before touching anything.
dev-prototype
Builds a throwaway experiment to answer a hard technical question before committing to a full implementation. Use when it's unclear whether something is technically feasible — the prototype is always deleted once the question is answered, so nothing ships until there's real confidence it will work.
dev-tdd
Enforces test-first development — write a failing test first, then write only the minimum code needed to make it pass, then clean up. This catches bugs early and ensures every feature ships with test coverage. Use when building anything that needs to be reliable.
devops-git-guardrails
Adds safety rules that block the most dangerous git commands — force-pushing to main, accidentally staging all files at once, and bypassing commit checks. Prevents the most common ways work gets accidentally lost or overwritten. Use when the operator says "git hooks", "protect main", or "guardrails", or after an incident involving force-push or lost work.
qa-triage
Classifies every incoming bug by severity (P0 = site completely down, P1 = major feature broken, P2 = significant issue, P3 = minor cosmetic), scores its priority, and routes it to the right person. Always run this first for any new bug report — nothing gets worked on without being triaged. Output goes to docs/qa/ and the project status dashboard. Use for every new bug report, or when the operator says "triage", "classify this bug", or "prioritise this bug".
il-project
This skill should be used when the operator says "il project", "new project", "scaffold a project", "create infinite leverage project", "init new project", "start new client project", or "bootstrap project folder". Scaffolds a brand-new project directory from the canonical `templates/project-scaffold/` in `talentedgeai/infinite-leverage`, substitutes placeholders, wires the agent team into `.claude/`, seeds `docs/product/` (product.md, epics.md, epic-status.md) from any rich description the operator provides, seeds `docs/brand/` styling from a chosen or random getdesign.md reference, initializes git, and prints next steps. All operations are inline — no bundled scripts.
designer-weekly
Tuesdays at 9:03 AM — finds the most recently written blog.md with no hero image yet, generates the hero image via the generate-image skill, saves to content/topics/{slug}/hero.webp.
pm-weekly-contribution
Mondays at 7:30 AM — runs pm-contribution-sync across all projects to refresh human-token and Claude-token data for the prior week, then runs pm-hub-report to update the centralized cross-project dashboard.
writer-weekly
Mondays at 9:03 AM — picks the oldest brief.md with no blog.md yet, writes the full blog post following the Neil Patel self-critique checklist, saves to content/topics/{slug}/blog.md.
dev-diagnose
Works through a confusing bug with a six-step scientific process: reproduce it reliably, narrow it down to the smallest case, form a theory about the cause, test the theory, apply the fix, then verify the fix holds under all conditions. Use when a bug is hard to pin down.
dev-grill
Stress-tests a plan by asking hard questions about what could go wrong — edge cases, missing assumptions, and failure modes. Use before starting any significant implementation to catch problems in the plan rather than in the code. Asks 10 or more adversarial questions and does not proceed until all are answered satisfactorily.
dev-zoom-out
Produces a plain-English summary of how a module or area of the codebase works — what it does, how its pieces fit together, and what to watch out for. Use when starting work in an unfamiliar part of the code or after a long break from an area.
devops-setup-pre-commit
Installs automatic code quality checks that run before every commit — catches style violations, formatting issues, and type errors locally before they ever reach GitHub or the live site. Run once per project during initial setup. Use when the operator says "pre-commit", "husky", or "lint-staged", or once during initial project setup.
global-caveman
Use when any agent session is running long and token costs are becoming a concern, or when the user says "caveman mode", "compress", "go minimal", or "token budget". Switches the agent to extreme brevity — no prose, only primitives. Available to all agents.
pm-grill-with-docs
Use when a plan or spec has been drafted and needs validation against existing project documentation before approval. Triggered when the PM says "grill with docs", "validate this plan", "check this against the project", or before approving any plan that touches existing epics. Interrogates the plan against project-status.html, epics.md, and epic-status.md.
create-agent
Runs a structured interview to design a new Claude Code subagent role from scratch. MUST be used when the user says 'create an agent', 'build an agent', 'I need an agent that...', 'set up a new agent role', or describes autonomous behaviour they want to delegate. Produces the complete agent package (persona, skills, evals) and installs the agent into .claude/agents/.
create-local-routine
Create a persistent scheduled routine using mcp__scheduled-tasks — stored on disk in ~/.claude/scheduled-tasks/, local timezone, no expiry, no minimum interval. Runs while Claude Desktop is open; catches up on next launch if the app was closed.
create-remote-routine
Use this skill whenever the user wants to create a scheduled routine, automated task, cron job, or recurring agent that sends outbound notifications, emails, messages, or HTTP requests to external services. Triggers on: "create a routine", "schedule a task", "send daily/weekly X", "automate sending", "set up a cron", "remind me via", "notify me when", or any request to periodically trigger an action in Lark, Slack, Discord, Telegram, Resend, SendGrid, Twilio, ntfy, Airtable, Notion, Make, Zapier, or similar services.
dev-feature-plan
Turns an approved feature spec into a concrete build plan and task checklist — validates the feature branch, resolves dev-layer findings, writes impl-plan.md (phases, decisions, risks, estimates) and tasks.md (dependency-ordered checklist with acceptance criteria). Self-contained: no other skills are invoked. Use when the operator says "plan this feature" or "how do we build this", or when pm-epic-writing hands off a completed spec.
devops-cicd
Creates a GitHub Actions pipeline that automatically runs quality checks on every pull request: checks code style, verifies TypeScript types, runs tests, and confirms the app builds successfully. This prevents broken code from reaching the main branch. Use when the operator says "ci/cd", "pipeline", "github actions", or a repo has no checks running on pull requests.
devops-ops
Monitors the health of the live site and CI/CD pipeline — checks deployment status, reads build and error logs, confirms all environment variables are correctly set. Also includes the step-by-step procedure to roll back a broken production deployment in under 60 seconds without writing any code. Use when the operator says "is the site up", "deployment failed", "check the logs", "roll back", or anything about production health.
github-flow
Complete GitHub workflow skill for clean, consistent branch-to-merge cycles. Use this skill whenever you need to: create a branch, make commits, push changes, open a pull request, merge a PR, or clean up after a merge. Also invoke proactively when the user says "commit this", "push my changes", "open a PR", "merge it", "clean up the branch", "I'm done with this feature", or any variation of git/GitHub operations. Agents are authorised to merge PRs — no human approval is required unless the skill explicitly flags a risk. This skill replaces ad-hoc git commands and ensures every change follows the same clean workflow.
marketing-strategist
Turns a client interview or business briefing into a complete marketing strategy — who the audience is, what messaging will resonate, which channels to focus on, and a 90-day content calendar to execute against. Run once at the start of a new project or campaign. Use when the operator says "marketing strategy", "new campaign", "client brief", or "content calendar", or at the start of a new project.
plan-protocol
Installs, upgrades, and diagnoses the Plan Protocol in any repository — the plan registry, the blast-radius guard, and the runtime-agnostic pre-push hook that stops undeclared mega-PRs. Use when setting up a new client project, when asked to add plan/scope enforcement, when `plans:*` or `plan.mjs` commands are missing or failing, when the pre-push guard is not firing, or when a fresh clone or worktree needs the protocol activated. Works in any stack — Node, Rails, Django, static — because the engine has no dependencies.
pm-analyze-split
Run immediately after speckit-analyze produces its findings table. Splits findings into PM layer (client-facing, business language) and Dev layer (routed to developer agent). HIGH PM-layer findings must be resolved before writing the epic entry. Called within pm-epic-writing.
pm-clarify-guard
Run immediately after speckit-clarify generates candidate questions. Filters out any technical questions before they reach the client — only business-level questions answerable by a non-technical owner are presented. Called within pm-epic-writing.
pm-contribution-sync
Fan-out skill: discovers every project under ~/code-projects/, runs scripts/team-hours.py for each, updates the #contributors table and #pulse chart in each project's docs/project-status.html, and writes a scripts/contribution-snapshot.json per project for hub aggregation. Respects the three-basis methodology and all editorial rules from docs/assessments/team-hours-methodology.md.
pm-documentation
Creates and maintains the single source of truth about the business — who the customers are, what problem is being solved, and what the product does. Run once after the client interview to create the file, then update it whenever the strategy changes. Everything the team builds is anchored to this document. Use when the operator says "update product.md" or "document the product", or right after pm-client-interview completes.
pm-epic-writing
Takes a new feature idea through the full discovery process — writes a structured spec, runs a business-level clarification interview, reviews the spec for gaps and conflicts, and turns it into a Dan Shipper epic the developer can build from with confidence. Self-contained: the whole spec pipeline (specify → clarify → analyze → epic) lives in this skill. Use when the operator says "write a new epic", "specify a feature", "add an epic for X", or after a client conversation produces a new idea. Output goes to .specify/features/{slug}/ and docs/product/epics.md.
pm-hub-report
Adaptive cross-project hub skill. Explores what centralized dashboard pattern the user already has, expands upon it, or creates ~/.claude/hub/hub.html as the canonical fallback. Aggregates contribution-snapshot.json from every project under ~/code-projects/ into one cross-project view showing per-project human tokens, Claude tokens billed, commits, and PRs. Additive — never overwrites existing hub structure.
designer-design-system
Fills and maintains the designer-owned half of the project's brand guide — colour palette, typography, and visual style — at docs/brand/style-guide.md. One brand identity, not a menu of themes. Leaves voice, vocabulary and content formats to the PM and writer. Use when the operator says "design system", "style guide", "brand colors", "visual identity", or when docs/brand/style-guide.md still has placeholder values in the palette, typography, or visual-style sections.
designer-image-generation
Generates one hero image using the Gemini AI image API, optimizes it to a compact web-friendly file (WebP, under 200 KB), and saves it alongside the blog post. Reads JSON image prompts from image-prompts.md created by the Writer. If image-prompts.md is missing, invokes the Writer to generate it first. If generation fails, saves the prompt so it can be used in any external image tool. Use when the operator says "generate the image", "hero image", "create a visual", or a finished post is missing its hero image.
designer-style-to-photo
Reads a blog post's tone and subject matter, picks a visual treatment within the project's single brand palette, and tunes the Writer's existing JSON image prompt so the generated visual is on-brand — filling in palette, style, and mood from the brand guide. Annotates the Writer's image-prompts.md; never authors a prompt from scratch. Use when the operator says "image prompt", "visual for this post", or a blog draft is finalized and its image prompt needs brand alignment before generation.
designer-ui-ux
Reviews a page or component against accessibility, responsive and usability standards — WCAG contrast, semantic markup, every interactive state, touch targets, and layout at four widths — and reports what fails with the fix. Use when the operator says "mockup", "wireframe", "ui design", "accessibility", "responsive", or asks "does this look right" about any page, and before any UI work is handed to QA.
email-marketer-nurture
Manages the full subscriber email lifecycle — drafts a welcome sequence for new subscribers, prepares weekly digest emails featuring the latest blog posts, and re-engages subscribers who have gone quiet. Every email is drafted for operator approval; this skill never sends. Uses Resend for individual transactional emails and Brevo for larger bulk campaigns. Owned by the writer agent. Use when the operator says "email campaign", "newsletter", "welcome sequence", "send to subscribers", or "re-engage subscribers".
pm-project-status
Builds and updates the project status dashboard — a single HTML page showing which features are planned, in progress, and shipped, plus a PDF companion regenerated on every update. Includes a build log, key metrics, and links to all important docs. The operator uses this to understand where the project stands at a glance. Use when the operator says "project status", "update the dashboard", or "where are we", and after any feature ships.
web-publisher-publish
Takes a finished blog post all the way to production — writes the App Router page, updates the blog index, commits on a publish branch, opens a PR, and confirms the Vercel deployment is healthy. Never commits or pushes on main. The post is not done until the build is green. Owned by the developer agent. Use when the operator says "publish", "push the post live", or "build the page", or a finished post + hero image are waiting.
writer-quality-critique
Critiques a draft blog post through the lens of Dan Shipper (every.to) — evaluating intellectual depth, original point of view, narrative pull, and conceptual originality. Returns a scored critique with specific revision instructions. The writer applies all feedback before passing to the Neil Patel SEO critique. Use on every draft before SEO review (mandatory gate), or when the operator says "critique this draft" or "is this post good".
writer-seo-content
Writes one complete, SEO-optimized blog post per run based on an approved brief. Applies a rigorous self-critique pass before finalizing — checking the hook, structure, evidence, readability, and call to action. Always produces the post file and a visual brief for the Designer to generate the hero image. Use when the operator says "write a post", "blog post", or "draft content", or the content calendar has an unwritten topic due.
dev-github-hygiene
Ensures every change follows clean practices: proper branches, clear commit messages, documented env vars in .env.local (never create .env.example files), and engineering notes for each change. Use at the start of any feature to keep the repository organized and auditable.
dev-karpathy
Applies a set of proven engineering principles that keep code simple, readable, and easy to test: always plan before coding, build in small verifiable steps, and prefer the simplest solution that works. Named after Andrej Karpathy's approach to clear, thoughtful engineering. Apply before and during any implementation.
dev-multi-agent
Breaks a large, complex task into independent pieces and works on multiple parts simultaneously, then brings the results together. Use for big features or refactors that touch many files — it dramatically reduces the time needed for large tasks.
dev-planning
Reads the current project status, open features, and known bugs, then creates a prioritized list of what to build today. Run at the start of each working session to get a clear, approved plan before writing any code.
dev-qa-delegation
Hands off completed work to the QA agent, tracks and fixes the bug reports that come back, manages the pull request review, and closes the loop after QA sign-off. This is the final step of the development workflow — use it when implementation is done.
pm-client-interview
Guides a structured two-round interview with the business owner to understand their goals, their customers, and what success looks like. Run once at the very start of a new project — this conversation becomes the foundation that everything else is built on. Output feeds directly into the product strategy document. Use when the operator says "client interview", "kick off the project", or "understand the business", or docs/product/product.md is still a placeholder.
pm-constitution-sync
Creates or updates the project constitution (versioned principles, KPIs, governance) at .specify/memory/constitution.md and syncs it to docs/product/constitution.md where every agent can find it. Use when the operator says "constitution", "project principles", "sync the constitution", or once during project setup after the client interview.
pm-to-issues
Breaks an approved spec or a generated tasks.md into dependency-ordered GitHub Issues via gh. Each issue is a vertical slice delivering user-visible value. Use when the operator says "to issues", "create issues", or "break this into tickets", or after a plan has been approved.
il-doctor
Check the Infinite Leverage plugin setup — prerequisites, repo context, and project scaffold health. Use when someone says "il doctor", "check my setup", "is infinite leverage working", "verify my install", or after installing/updating the plugin.
infiniteleverage-project
This skill should be used when the operator says "new project", "scaffold a project", "create infinite leverage project", "init new project", "start new client project", or "bootstrap project folder". Scaffolds a brand-new project directory from the canonical `templates/project-scaffold/` in `talentedgeai/infiniteleverage-8-agents-template`, substitutes placeholders, wires the 8-agent team into `.claude/`, initializes git, and prints next steps. All operations are inline — no bundled scripts.
infiniteleverage-validate
Validate that effort tracking is set up correctly for this contributor and repo. Use this skill when someone says "validate effort tracking", "am I being tracked", "check my telemetry setup", "is my Claude usage logged", "verify telemetry", "check if I'm logging hours", "am I capturing effort", or "is effort tracking working".
infiniteleverage-patch
Health-check local Claude Code setup, then fetch latest agent templates from GitHub and apply updates. Safe on Mac Mini or laptop.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.