Tool #8 · What we build with · How we work

Primary AI Coding AgentClaude Code (Opus 5.5)

Our daily-driver coding agent across 865+ human sessions and the majority of client commits—always paired with a second review gate.

Last reviewed: 2026-10-22 · Status: How we work

Book a 15-minute fit call Back to all 12 tools in The NexAI Stack
Full stack receipt#8 · How we work · reviewed 2026-10-22
#8Primary AI Coding Agent

Claude Code (Opus 5.5)

How we work
1. What we use it for: We use Claude Code (running Claude Opus 5.5 / Opus 4.6) as our daily-driver engineering agent to plan, write, test, and audit code across client mobile apps, websites, and backend services.
2. Client use case · RiseSyncState: used
Problem

Taking over an inherited 102-commit Expo + Fastify + Airtable app and reaching TestFlight in five days required rapid implementation without letting AI-authored bugs reach production.

What the tool did

Co-authored 115 of our 172 commits since 2026-10-17, while a second agent review pass caught a fail-open moderation bug before the first Milestone 1 production deploy.

Result

Milestone 1 merged and deployed to Railway on 2026-10-22 with 72 passing tests, and TestFlight build 9 submitted the same night.

2. Client use case · A print and signage company, a members' dinner club, and a nonprofit education centerState: used
Problem

Delivering 27-to-49-page site rebuilds and maintaining a full Angular/Express platform on fixed-price scopes requires high commit velocity backed by verifiable browser and test gates.

What the tool did

Co-authored 58 of 86 commits on a print and signage company’s site, 133 of 308 commits on a members’ dinner club’s site (plus independent visual review), and 205+ commits on a nonprofit education center’s platform.

Result

Used in 865 of 873 human desktop coding sessions since 2026-07-24 (305 of 324 recent sessions on Opus 5.5 across 171,000+ messages), plus delivered a $2,500 read-only production-readiness audit for a finance-app client in October 2026.

3. What it costs us or the client

$200/month on the Claude Max 20x subscription for interactive desktop and CLI builds, plus metered Anthropic API spend when headless workflows call the API directly (as of 2026-10-22).

4. Setup notes (how it is wired in)

Every repo carries a checked-in `AGENTS.md` / `CLAUDE.md` contract defining pinned runtimes, test gates, and confidentiality rules; agents work in isolated git worktrees and stamp `Co-Authored-By` trailers on every commit.

5. What broke & how it was handled

On RiseSync on 2026-10-19 (commit `9e58bdd`), Claude Code wrote `backend/src/services/moderation.ts` with a fail-open defect: if Airtable errored before the ban or hide lists loaded right after a deploy, both lists were treated as empty—meaning a banned user could sign in and opted-out posts would appear in galleries. A second agent review pass flagged it on 2026-10-22 at 14:30 UTC; within 3 minutes (commit `39266cb`, PR #18) we added a retry, a `503 MODERATION_UNAVAILABLE` fail-closed response, and 3 regression tests before the code ever ran in production.

6. What we compared it against & why we picked it

Compared against Cursor (which we retired in August 2025 after 4 commits on the nonprofit education center repo), Gemini CLI (retired in March 2026), OpenAI Codex (which we run alongside Claude Code for parallel branches and independent reviews), and secondary headless lanes (ZCode/GLM, Kimi, Antigravity, Qwen). We picked Claude Code with Opus as our primary builder because it holds deep repo context and follows strict multi-file contracts.

7. Last reviewed: 2026-10-228. Video / tutorial: Featured in Episode 1 (RiseSync takeover) and Episode 4 ("How We Run Claude Code + Codex + OpenBrain Across Client Builds")
All stack tools
← #7: Expo EAS + TestFlight#9: OpenAI Codex →
How an engagement runsthree steps, in order

Three separate steps. Stop after any of them.

Step 1 is a free 15-minute fit call. Step 2 is the AI Staff Assessment, $500, a written assessment you keep. Step 3 is your first hire at a published price, or you take the assessment and stop.

Step 1

Fit call

$0 · 15 min
  • tell me which job eats your week: the phone, the inbox, marketing or support
  • I tell you whether an AI hire fits, and whether I am the right person for it
  • you leave with the first hire I would start with
  • a slide deck

video call · same week

book it
Step 2

AI Staff Assessment

$500
  • a 60-minute video call about how your business runs
  • my written assessment the same day: whether an AI hire would pay for itself, the job I would hand off first, what to settle first, and the fixed price for that first hire
  • not taken off the build, and no separate quoting call: the assessment names the package at its published price
  • proposals written before anyone learned your business

paid when you book · not taken off the build

start here
Step 3

Your first hire

fixed price
  • Front Desk Hire or Brain + First Hire, at the published price
  • built in 2 to 6 weeks, then 30 days of tuning
  • or take the assessment and stop there
  • bench developers you never met

priced in the assessment

see the prices

Want Claude Code (Opus 5.5) wired into your business?

Bring the workflow or codebase to a free 15-minute fit call. We will tell you how we wire Claude Code (Opus 5.5) in production, what breaks first, and whether a $500 AI Staff Assessment fits your scope.

Book a 15-minute fit call
or email david@nexaiadvisors.com about Claude Code (Opus 5.5)