Service AI Engineering
Skills encode your conventions, runbooks, and tools so Claude operates as a domain expert. Your team gets a teammate that knows your codebase, your process, and your deploy script.
---
name: deploy-to-staging
description: Use when deploying any service to staging. Walks through the
pre-flight checklist, runs the deploy, verifies health checks, and posts
to #deploys. Triggers on phrases like "deploy to staging", "ship to
staging", "push staging".
---
# Deploy to Staging
## Steps
1. Run `make pre-deploy` and confirm all checks green
2. Tag the commit: `git tag staging-$(date +%Y%m%d-%H%M)`
3. Push: `make deploy ENV=staging`
4. Verify: curl staging health endpoint, expect 200
5. Post to #deploys with version + changelog
Why this matters
Out of the box, Claude can write a Rails controller. It cannot know that your team prefers Pundit over CanCanCan, that your background jobs use Sidekiq with specific retry semantics, or that your deploy runs through your custom CLI. Skills are how you teach Claude that. Done right, your team gets a senior engineer that already knows your stack.
What we build
Skills designed for natural-language invocation, with progressive context loading, real evals, and the distribution your team needs to install them in seconds.
01
Skill names and descriptions designed so Claude routes to them when your team types natural language. No "what skill do I run for this?" lookup. Trigger phrases match how your team actually talks.
Skill invocation rate above 80 percent of relevant requests.
02
Skills load lightweight context first, then deeper detail only when needed. Your team gets the right information without burning the entire context window on one task.
Average skill token cost stays under 2,000 tokens.
03
Every skill ships in a git repo. Pull requests review skill changes the same way your team reviews code. No more "who changed the prompt?"
Skill changes follow your existing change management.
04
Real cases, expected outputs, automated runs in CI. A skill change that breaks a real workflow blocks merge before it ships.
Skill regression rate falls to near zero.
05
Plugin packaging for Claude Code, marketplace listings, install one-liners, runbooks. Your team installs skills in seconds, not by hand-copying files.
Adoption above 70 percent of target users in week one.
06
Each skill ships with a one-page reference your team can read without knowing how the skill works internally. Your engineers can extend the skill without consulting us.
Your team can ship new skills after week six.
80%+
of relevant requests trigger the right custom skill across our shipped suites
Measured by skill invocation logs. Public methodology on request.
Skill structure
Markdown frontmatter for routing, body for instructions, optional referenced files for deeper context. Your engineers can read every skill we ship in five minutes.
# .claude/skills/code-review-rails/SKILL.md
---
name: code-review-rails
description: Use when reviewing Ruby on Rails pull requests. Checks ActiveRecord
N+1 queries, strong parameter coverage, missing scopes, security
patterns from the Rails guide, and team-specific conventions in
rails-conventions.md. Trigger on "review this Rails PR", "code review",
"PR check", "audit this controller".
---
# Code Review: Rails
Read `rails-conventions.md` for our team rules.
## Phase 1: Quick Scan
Read every changed file. List every public method, every callback, every
ActiveRecord query. Note anything that looks suspicious for the next phase.
## Phase 2: Deep Checks
For each query, confirm includes/preload covers all loaded associations.
For each controller action, confirm strong params match the model attributes.
For each new background job, confirm idempotency and retry behavior.
Process
01
One week. We shadow your team, identify the workflows worth encoding as skills, design the skill catalog, and lock the trigger phrases. You approve the catalog before any skill is written.
Fixed scope, fixed price.
02
Two to three weeks. Skills written, evals wired, packaging built. Your team installs and uses them in week two for real-world feedback.
Skills used in real work by week two.
03
One week. Documentation, maintenance runbook, training session for your team to extend skills themselves. We step back when your team can ship the next skill.
Your team owns the suite at the end.
Common questions
A skill is a packaged unit of context (prompt, instructions, examples, even scripts) that Claude loads on demand when the user asks for something the skill knows how to handle. Think of it as a function the model calls into, except the function is markdown your team can read.
Two to four weeks for a focused skill suite (5 to 10 skills covering one workflow area). Discovery and skill design takes one week, build and evals run two to three weeks. Faster if your conventions are already documented.
Anywhere a Claude Agent SDK app loads them: Claude Code, Claude Desktop with the Agent SDK, custom internal tools, or Claude API calls in your own services. We package skills so they install identically across all of them.
Yes. Skills can use the same MCP servers, Bash commands, and tools the parent agent uses. We typically pair a custom skill suite with a custom MCP server when your team needs to act on internal data, not just reason about it.
Eval suite in CI catches regressions. A monthly review process flags skills that have not been invoked in 30 days. We hand off the maintenance runbook so your team owns updates after launch.
Custom skill projects are scope-dependent for a focused suite of 5 to 10 skills. Larger skill libraries with custom evals and distribution are scoped after discovery. Discovery call is free.
Ready to teach Claude your conventions?
Discovery call is free. Fixed-price quote within 48 hours. NDA on request.