Bambu Skills

A curated collection of Agent Skills for AI coding agents, built for the way Bambu developers work.

---

What are skills?

Agent Skills are packaged instructions that extend the capabilities of AI coding agents (such as Claude Code). Each skill teaches the agent how to perform a specific task following Bambu's conventions — so that generated code, tests, and docs match what the rest of the team would write by hand.

An agent loads a skill automatically when it detects a relevant task. Skills override generic, framework-default advice whenever they conflict with our house style.

Installation

Install the entire collection:

npx skills add Bambu-Developers/skills

Or install an individual skill by name:

npx skills add Bambu-Developers/skills/bambu-e2e-test-matrix
npx skills add Bambu-Developers/skills/bambu-nest-rules
npx skills add Bambu-Developers/skills/bambu-nest-test
npx skills add Bambu-Developers/skills/bambu-readme-generator
npx skills add Bambu-Developers/skills/bambu-terraform-aws

Once installed, the agent will leverage each skill automatically when a matching task comes up — no manual invocation required.

bambu-e2e-test-matrix needs one extra setup step. Unlike the other skills, it drives a real browser and records video through the Playwright MCP, so it requires a one-time MCP configuration. After installing, run its bundled setup script once (it pins @playwright/mcp@0.0.68 with --save-video and optionally registers its sub-agents):

bash bambu-e2e-test-matrix/scripts/setup.sh

Then verify with /mcp that playwright is connected. See bambu-e2e-test-matrix/README.md for the manual equivalent and the optional Figma/SAST tooling.

Available skills

Skill	What it does
bambu-e2e-test-matrix	Generates and runs manual E2E test matrices (flow, usability, visual, accessibility, edge cases) against any web app via Playwright MCP — batch execution with continuous video evidence, optional Figma diffs and SAST security analysis, and a final scoring/ranking.
bambu-nest-rules	Project-specific conventions for our NestJS + Prisma monorepo — dynamic-module libs, Secrets Manager, typed envs, i18n, error handling, DI, and thin controllers.
bambu-nest-test	Canonical unit-testing patterns for our NestJS + Prisma monorepo — DTO, service, controller, and module tests.
bambu-readme-generator	Regenerates a project's root README.md by autodiscovering its real state — language, layout, and scripts.
bambu-terraform-aws	Reusable, project-agnostic conventions for generating, modifying, and reviewing Terraform infrastructure on AWS — modules, environments, networking, security groups, tagging, and the interchangeable compute layer.

bambu-e2e-test-matrix

Generic, self-contained skill for manual E2E QA on any web application. It runs in two phases: first it explores the site (and optionally the source repo, read-only) and proposes a test matrix (matriz-pruebas.csv) covering functional, flow, usability, visual, accessibility, and edge/negative cases — then, only after you approve it, it executes. Execution is delegated to sub-agents batch by batch (one module per e2e-runner; the orchestrator only coordinates, never navigates), records one continuous video per module as evidence, checkpoints progress to ESTADO-CORRIDA.md so a run can resume where it left off, and enforces a completeness gate so no approved case is silently left pending. It can optionally diff screens against a Figma design and run static SAST security analysis (OWASP Top 10, CWE Top 25, hardcoded secrets, dependency CVEs), and closes with a 0–100 score, letter grade, and per-module ranking. Everything is saved locally; it never modifies the code of the app under test. Requires the Playwright MCP (see the extra setup step under Installation). Load it when you ask for an "E2E test matrix", "flow/usability testing", "test a site", "video test evidence", "compare the UI against Figma", or "SAST/security analysis of the code".

bambu-nest-rules

Project-specific conventions for the Bambu NestJS monorepo. Covers forRoot/forRootAsync dynamic modules in libs, AWS Secrets Manager integration, typed envs via config/env.config.ts, DTOs with i18n validation, I18nModule setup in apps, service-layer error handling with I18nService.t(...), PrismaService usage, constructor-based dependency injection, and thin controllers. Load it when writing, reviewing, or refactoring any code under apps/<service>/ or libs/<lib>/. Pairs with bambu-nest-test so tests mirror the implementation conventions.

bambu-nest-test

Unit-testing patterns for the Bambu NestJS monorepo. Covers plainToInstance + validate for DTOs, Test.createTestingModule with mocked PrismaService and I18nService for services, thin-controller delegation assertions, and module DI-wiring verification. Load it when writing or reviewing tests under apps/<service>/src/ or libs/<lib>/src/.

bambu-readme-generator

Language- and framework-agnostic README generator. It inspects manifests (package.json, pyproject.toml, go.mod, Cargo.toml, …), maps the repo layout, extracts runnable scripts, and renders a canonical README from a template — treating the existing one as a stale snapshot. Load it when you ask to "generate", "update", or "refresh" the project README.

bambu-terraform-aws

Reusable, project-agnostic conventions for Terraform on AWS. Covers module and environment structure, naming, two-layer tagging, for_each/count iteration and deterministic outputs, a 3-layer VPC with private data tiers, one-SG-per-component security groups with SG-to-SG references, a non-negotiable Well-Architected security baseline (private networking, encryption, secrets, least-privilege IAM), and an interchangeable application/compute layer (Lambda / Fargate / EC2 / EKS) on a common base. Load it when creating or reviewing modules under modules/ or environments under environments/, wiring the VPC, subnets, or security groups, or choosing/switching the compute layer.

Repository layout

skills/
├── bambu-e2e-test-matrix/    # manual E2E QA (Playwright MCP)
│   ├── SKILL.md
│   ├── scripts/              # setup.sh — one-time Playwright MCP config
│   ├── references/           # browser/visual/a11y/SAST guides + agents/ profiles
│   └── assets/templates/     # matriz-pruebas.csv · ESTADO-CORRIDA.md · reporte.md
├── bambu-nest-rules/         # NestJS project conventions
│   ├── SKILL.md
│   └── rules/                # progressively-disclosed rule files
├── bambu-nest-test/          # NestJS unit-testing patterns
│   ├── SKILL.md
│   └── rules/                # progressively-disclosed rule files
├── bambu-readme-generator/   # README generator
│   ├── SKILL.md
│   └── templates/            # skeleton + section recipes
├── bambu-terraform-aws/      # Terraform/AWS infrastructure conventions
│   ├── SKILL.md
│   └── rules/                # progressively-disclosed rule files
└── CLAUDE.md                 # guidance for agents working in THIS repo

Each skill is a self-contained directory whose SKILL.md carries YAML frontmatter (name, description) plus instructions for the agent. Supporting rules/ and templates/ files are loaded on demand.

Contributing a new skill

1. Create a directory named after the skill (kebab-case).
2. Add a SKILL.md with name (must match the directory) and a description that enumerates concrete trigger phrases — the description is the only thing the agent sees when deciding whether to load the skill.
3. Keep SKILL.md as an index; push detail into numbered rules/*.md or templates/*.md files.
4. Write instructions for a future agent, not end-user docs: when to load, how to apply step by step, and the hard constraints.

See CLAUDE.md for the full authoring conventions.

License

MIT © Bambu Tech Services