agent-resource-kit

startup-data-kit is a set of agent skills for Claude Code, Cursor and Codex. They guide an agent through setting up a Bruin pipeline for revenue, marketing, product, finance or AI coding spend data, and through answering questions from the tables it builds.

npx skills add bruin-data/agent-resource-kit

Try it with sample data

This runs the ai-coding-usage template on DuckDB with generated Claude Code and Cursor usage data. It needs no credentials and no warehouse, only the Bruin CLI and git. Run it in an empty folder outside any git repository.

# 1. Get the template and the sample data
git clone --depth 1 https://github.com/bruin-data/agent-resource-kit
bruin init ai-coding-usage ai-coding-demo

# 2. Swap the Anthropic and Cursor API assets for the fake CSVs
cp agent-resource-kit/resources/startup-data-kit/ai-coding-spend/sample-data/*.{asset.yml,csv} \
   bruin/ai-coding-demo/assets/raw/

# 3. Build every table on DuckDB, then query one
cd bruin
bruin run ai-coding-demo --full-refresh --workers 1 --start-date 2026-09-14 --end-date 2026-09-27
bruin query -c duckdb-default -q "SELECT strftime(date_trunc('week', usage_date), '%Y-%m-%d') AS week, platform, CAST(SUM(total_tokens) AS BIGINT) AS tokens, ROUND(SUM(estimated_cost_usd), 2) AS est_cost_usd FROM marts.ai_coding_usage_by_user_day GROUP BY ALL ORDER BY ALL"

Then, with the skill installed, ask your agent why estimated Claude Code cost rose in the second week. sample-data/ describes the data and explains each flag. --full-refresh drops and rebuilds every table, and is used here only because the demo database is new.

What is here

Open-source skills for AI agents. Each resource is a SKILL.md file in the format that Claude Code, Cursor, Codex and other agents load automatically. Copy a folder into a project and the agent follows its instructions. There are two kinds:

• Pointers to existing tools, so an agent uses the maintained tool instead of writing its own.
• Task skills for decisions that no existing tool covers.

Join the Slack community for questions, help with a setup, and feedback. If a skill led your agent in the wrong direction, report it there or in an issue.

Resources

Setup

Resource	Status	Risk	Use it when
bruin-agent	experimental	read-only	Installing the Bruin CLI, creating a project, choosing a template, connecting a source, registering the Bruin MCP server, installing Bruin's agent skills

startup-data-kit: an agent works with you to set up and customise a Bruin template for your startup, then answers questions from it. One domain per folder:

Resource	Status	Risk	Use it when
revenue-analytics	experimental	approval-required	MRR, churn, expansion, failed payments, cash collection. Stripe, Chargebee, Paddle
marketing-analytics	experimental	approval-required	CAC, ROAS, attribution, campaigns, CRM pipeline. Google Ads, Meta, HubSpot, Klaviyo
product-analytics	experimental	approval-required	Activation, feature adoption, retention, usage-based churn signals. PostHog, Mixpanel, Amplitude
web-analytics	experimental	approval-required	Traffic, landing pages, channels, funnel drop-off. GA4, Search Console
ai-search-visibility	experimental	approval-required	SEO, GEO and AEO. Search Console, AI crawler traffic, whether assistants mention you
finance-accounting	experimental	approval-required	Receivables and aging, invoiced vs collected, customer concentration, vendor balances. QuickBooks Online
support-analytics	experimental	approval-required	Ticket volume, response and resolution time, CSAT, backlog, support load per customer. Gorgias, Zendesk, Intercom
ai-coding-spend	experimental	approval-required	AI coding tool adoption, usage and estimated cost by team, model and platform. Claude Code, Cursor
sheets-and-notion	experimental	approval-required	Hand-maintained plans, targets, budgets and mappings, joined to the other domains. Google Sheets, Notion

data-engineering-skills: keeping pipelines healthy. Install Bruin's seven maintained skills, then add the orchestration layer they do not ship:

Resource	Status	Risk	Use it when
bruin-builtin-skills	experimental	read-only	Working out which maintained Bruin skill to install and reach for
pipeline-triage	experimental	read-only	Something broke and you need to work out what, and route it
anomaly-investigate	experimental	read-only	A metric moved but nothing failed
maintenance-pr	experimental	approval-required	Turning a proposed fix into a reviewed pull request
pipeline-backfill	experimental	approval-required	Rerunning a date range safely after a fix
pipeline-report	experimental	approval-required	Telling a human what happened

coding-agent-workflows: repeatable work on a codebase.

Resource	Status	Risk	Use it when
generate-changelog	experimental	read-only	You need customer-facing release notes from git history
record-terminal-demo	experimental	approval-required	You need a terminal demo of a CLI recorded with VHS, rendered and visually checked

tools: sample scripts an agent can shell out to. These ship code you run yourself, with your own credentials.

Resource	Status	Risk	Use it when
attio-cli	experimental	approval-required	Reading or updating records, lists and attributes in an Attio CRM
blog-image-generator	experimental	approval-required	Finding a licence-friendly Unsplash photo and turning it into a cropped, treated cover image
linkedin-enrichment	experimental	approval-required	Pulling public LinkedIn company, person, post or job data through paid Apify actors

Everything is experimental today.

Risk is the worst a resource can do when followed as written. read-only reads and produces local files. approval-required can write, run or send something, and every such action is gated on your explicit approval. SECURITY.md explains what the labels do and do not guarantee.

Use it

With skills.sh, for Claude Code, Cursor, Codex and the other agents it supports. It lists every skill here and asks which to install and for which agents.

npx skills add bruin-data/agent-resource-kit
npx skills add bruin-data/agent-resource-kit --skill ai-coding-spend   # just one
npx skills add bruin-data/agent-resource-kit --list                    # look first

As a Claude Code plugin. Each group is a plugin: bruin-agent, startup-data-kit, data-engineering-skills, coding-agent-workflows and tools. Plugin skills are namespaced, such as /startup-data-kit:ai-coding-spend.

/plugin marketplace add bruin-data/agent-resource-kit
/plugin install startup-data-kit@bruin-agent-resource-kit

Or copy the folder. Agents load skills from these folders without further configuration.

git clone https://github.com/bruin-data/agent-resource-kit
cp -R agent-resource-kit/resources/bruin-agent ~/my-project/.claude/skills/
cp -R agent-resource-kit/resources/startup-data-kit/revenue-analytics ~/my-project/.claude/skills/

.claude/skills/ for Claude Code, .agents/skills/ for several others. Bruin's own bruin ai skills uses whichever of the two the project already has, and links both if both exist (commands/ai-skills). Skills are independent; take only what you need.

Or hand your agent the link.

Read https://github.com/bruin-data/agent-resource-kit and use whatever fits:
I need to work out why our MRR stopped growing last quarter.

It reads the index, then the one skill it needs.

Tested

TODO: results pending. The comparison runs separately and fills in this table.

Question	Without skill	With skill
TODO	TODO	TODO
TODO	TODO	TODO
TODO	TODO	TODO

Upstream documentation is the source of truth

Every file here is a snapshot of how something worked the day it was written. Where a resource disagrees with the tool's own documentation or its --help output, upstream is right and the file here is stale. Each resource says which version and date it was written against, and instructs the agent to check upstream and tell you when it finds a difference.

These resources say which tool to use and what to be careful about. They do not duplicate the tools' documentation.

What the skills cover

The main risk when an agent works with business data is not an error message. It is a plausible number calculated from the wrong definition.

Each skill covers the definitions that change the answer, the known pitfalls of that data source, what the source cannot tell you, and which actions need a person's approval. Skills point to maintained tools instead of reimplementing them, and link to upstream documentation instead of copying it, so that they do not go out of date.

Credentials

Nothing here will ask you to paste a credential into a chat, and every skill tells the agent not to.

A key sent in conversation sits in the transcript, likely in provider logs, and in the agent's context for the rest of the session where it can resurface in a summary or a generated file. You cannot revoke it from a transcript, only from the provider. A key passed as a command flag is in your shell history and the process list.

Use browser OAuth, an interactive local prompt such as bruin connections add, an environment variable reference, or your secret manager. Ask for read-only scopes. A scoped credential enforces limits; instructions in a file do not.

This repository contains no credentials and no real business data.

Contributing

Contributions of resources you have used in practice are welcome. CONTRIBUTING.md describes the format, the labels, and what gets rejected.

./tests/check.py

No dependencies. It checks frontmatter, status labels, upstream notes and their dates, index and plugin manifest completeness, local links, sample data that uses only reserved example domains, and files that must never be committed. CI runs it on every change, scans for credentials with gitleaks, and checks external links weekly to catch broken upstream links.

Provided as-is

Example code, provided as-is. This repository is not part of the Bruin platform, not a supported Bruin product, and not covered by any support agreement or SLA. It is documentation and sample scripts: it stores no customer data, holds no credentials, and has no access to any production system. Scripts run on your own machine, under your own credentials, configured by you. See LICENSE for the warranty disclaimer.

Licence

MIT. See LICENSE.