skills AGENTS.md

AGENTS.md

Guidance for AI coding agents working with this repository.

Repository Overview

A collection of skills for AI agents following the Agent Skills Specification.

Structure

skills/
  {skill-name}/
    SKILL.md
    references/
      {topic}.md

scripts/
  score-skills.py
  scorer/
  tests/

Creating a New Skill

Naming Convention

Use an imperative verb, lowercase, and hyphens only:

• analyze-data ✓
• analyzing-data ✗

SKILL.md Format

---
name: {skill-name}
description: {Third-person description with trigger keywords}
---

# {Skill Title}

## Quick Start
{Immediate actionable example}

## When to Use This Skill
{Bullet points with trigger conditions}

## Common Pitfalls
{5-10 mistakes to avoid}

## References
{Links to references/ files}

Best Practices

• Keep SKILL.md under 500 lines
• Move details to references/
• Third-person descriptions ("Analyzes..." not "I help you...")
• Include trigger keywords
• References one level deep only

Commands

• /altertable:ask <query> -- routes user queries to the best skill (see skills/ask/SKILL.md)

Available Skills

<available_skills> analyze-funnels Builds conversion funnels over ordered steps. Use for user journeys, drop-off, onboarding, checkout, or multi-step flows. Returns a funnel insight. analyze-insights Explains what an existing insight or chart shows. Use when asked what a visualization or graph means, or to read out its outliers. Does not run a query. analyze-web-traffic Analyzes website traffic (pageviews, sessions, referrers, landing pages). Use for site visits, traffic sources, bounce, or visitor behavior. ask Routes user queries to the best-fit Altertable skill. Use when unsure which Altertable skill applies to a request. build-segments Compares event metrics across cohorts using filters, breakdowns, and dimensions. Use to define an audience or compare groups such as free versus paid. Returns a segmentation insight. configure-tasks Schedules recurring AI tasks over insights and dashboards. Use for alerts, cron-style monitoring, or anomaly and forecast checks that run on their own. create-discoveries Reports a change, anomaly, root cause, recommendation, or warning to users. Use when analysis produces something worth telling someone. Returns a discovery that enters review. create-insights Drafts, renders, and saves insights of every type (SQL, semantic, segmentation, funnel, retention). Use to build, preview, or share a visualization. Returns a saved insight. decide-actions Decision matrices for picking insight types (funnel, retention, semantic, segmentation, SQL), task types, and discovery actions. Use when choosing types or whether to create, update, or skip discoveries. evaluate-skills Evaluates and authors agent skills against the Agent Skills spec. Use when reviewing, writing, or refactoring a SKILL.md, or asking about structure, frontmatter, or naming. explore-data Inspects catalogs, schemas, tables, columns, semantic models, measures, and dimensions. Use to find what data exists or a table's columns. Reads metadata, runs no query. forecast-timeseries Runs on-demand statistics over a time series to spot outliers and project values. Use for whether a spike is normal, or what a metric reaches next week. instrument-product-analytics Adds Altertable product analytics to an application (event tracking, user identification, traits, consent, session reset, aliasing). Use when writing instrumentation code. manage-discoveries Reviews, approves, and rejects existing discoveries. Use for the approval queue, discovery states, or user feedback on a discovery. query-lakehouse Writes, validates, optimizes, and runs DuckDB SQL against the Altertable lakehouse. Use when answering a question requires executing a query, joining tables, or aggregating metrics. query-product-events Queries product events and identities with SQL (event counts, properties, user activity, traits). Use to answer a question about tracked behavior or confirm new tracking arrives. understand-platform Explains Altertable concepts and architecture. Use for what Altertable is, or how agents, discoveries, insights, memories, and dashboards relate. Concepts only. use-memory Stores and recalls agent memories and org knowledge between sessions. Use to remember a definition, threshold, or convention, or recall earlier context. Not user-facing. </available_skills>