loadouts / coding
Loadout
Coding Agent
Plan the work, ship the code, watch it in production — one loop.
5 servers6 tools1 write-capable
For: Engineers and founders whose agent works across the repo, the issue tracker, deploys, and product analytics.
01The kit · 5 servers
GitHub Repo & PRs
Read code, manage issues, open pull requests — the spine of an engineering agent.
Linear Issue tracking
Plan and track the work — create, triage, and update issues right in the loop, so code and plan stay in sync.
Vercel Deploys & logs
Ship it and watch it — inspect deployments, read build and runtime logs, and debug production, OAuth-scoped.
PostHog Product analytics
Query events, insights, and error tracking, so the agent measures what actually shipped — not what it hoped.
Ahrefs Search & SEO
Real keyword, SERP, and backlink data for when the work is web-facing and has to be found.
02Governance
The reads — code, issues, deploy logs, analytics — are the safe foundation of every good change. The writes are where consequences live: PRs get reviewed before merge, deploys are OAuth-scoped to exactly what you grant, and issue updates are cheap to undo. Arranged so the agent looks (repo, logs, data) before it leaps (PR, deploy).
5 read1 write
03Plays
Fix a bug, end to end
- Pick up the issue and read the repo (Linear, GitHub)
- Open the fix PR (GitHub)
- Verify the deploy and its logs, then close the issue (Vercel, Linear)
Ship & measure a feature
- Build against the repo, tracked in Linear (GitHub, Linear)
- Deploy and check build + runtime logs (Vercel)
- Watch adoption and errors in analytics (PostHog)
Build a page that ranks
- Research the keyword and SERP (Ahrefs)
- Build the page and ship it (GitHub, Vercel)
- Measure traffic and behavior (PostHog)
Run this loadout
Assemble Coding Agent in Loadouts — the servers wired up, the governance enforced, the plays ready to run.