The engineer who reads everything before touching anything

Glenn, also known as Mr Code, is the AI Engineering Agent at Landing Pad Digital. He runs on the VPS, reads the codebase, ships features, fixes bugs, writes tests, updates documentation, commits cleanly, and pushes to git. He works in disciplined sessions with explicit READ FIRST blocks, verifiable success criteria, and honest report-backs. Glenn is the reason a small team can run a serious production stack without hiring senior engineers.

Glenn - AI development teammate watercolour illustration

The situation before

The team's infrastructure was already significant: a Hostinger VPS in Jakarta running Chatwoot, Odoo, n8n, Nextcloud, Listmonk, Matomo, Uptime Kuma, plus the Solutions staging app and supporting services. Maintenance, feature work, and incident response were all handled by the founder personally. Anything that needed code (a new n8n workflow, a Next.js page, a database schema change, a Docker build, an SSL renewal) was a context-switch out of running the business and into the terminal.

The team did not need a full-time engineer. The work was bursty - some weeks needed a lot, some weeks needed nothing. A salaried engineer would be either overworked or underused, and would still take weeks to learn the stack. The team also did not want to depend on freelancers who would never quite know the codebase well enough to ship safely.

What the team needed was someone who could read the entire stack on demand, ship a defined task with verification, and never lose context between sessions. Someone who could be summoned for a specific job, do it cleanly, and step back.

How Glenn is set up

Roles deployed

Environment
Glenn lives inside Claude Code on the VPS. He has filesystem access to every repo (`landingpad-solutions-staging`, `stepify`, `wikis`, the Docker stack), `bash` for executing builds and tests, and read access to logs and database queries.
Engineering memory
every project Glenn works on has a `docs/dev/` directory containing four files he reads at the start of every session: `CURRENT-STATE.md` (what is working and broken), `VERIFIED.md` (what is confirmed working, do not re-investigate), `KNOWN-ISSUES.md` (gotchas that waste time), `SESSION-LOG.md` (what previous sessions tried). These files are version-controlled and updated by Glenn himself at the end of every session.
Skills library
at `.claude/skills/` Glenn has 24 reusable skill modules: `executing-plans` for task-by-task discipline, `verification-before-completion` for the "is this actually done" gate, `systematic-debugging` for root-cause work, `senior-frontend` and `senior-backend` for stack-specific patterns, `docker-ops` for builds, `git-commit-helper` for clean commits, plus 17 more.
Agent suite
at `.claude/agents/` Glenn can call subagents when a session calls for it: `debug-detective`, `expert-nextjs-developer`, `testing-evidence-collector`, `testing-reality-checker`, `audit-project`, `research-core`, and others.
Session prompts
every Glenn session begins with a structured prompt that defines the scope, names the files to read, lists rules ("staging only", "no `--no-cache`", "no em dashes"), enumerates tasks in order, and ends with a SUCCESS CRITERIA checklist and a REPORT BACK template. The prompt is the contract.
Verification first
Glenn does not declare a task done until `verification-before-completion` confirms it actually works (curl returns 200, build passes, DB row lands, container is healthy). The "compiled but not verified" failure mode that plagued earlier sessions is now gated.

"Pulled origin/main, read the four memory files, scanned the affected modules. The B2 wiki utility extension landed clean: four new functions added, the type extended, empty-directory handling tested against the empty `wiki/case-studies/` directory. Twelve files changed, single commit, pushed. Engineering memory updated. Anything to flag: the gray-matter advisory is a transitive postcss issue from Next.js bundled deps, not gray-matter itself. Documented and left alone per plan policy."

Tier

tier-3

Department

operations

Status

Live

More case studies

Case study

The pipeline that does not leak

Iris is the AI Sales Agent at Landing Pad Digital.

Case study

Customer Success that does not sleep

Lanna is the AI Customer Success Agent at Landing Pad Digital.

Want one shaped for you?

Tell us your context. We'll come back within one business day with a representative shape, a likely tier, and an honest timeline.

Submit a Hiring Brief