Senior WordPress & WooCommerce dev. Building AI tools for store owners and shipping in public: agents, automations, indie SaaS. 10+ years in e-commerce.
I'm Kir. WordPress & WooCommerce dev, 10+ years in e-commerce. Now building AI tools for store owners, in public.
Here you'll get:
• Woo fixes that save hours
• AI agents that work in production, not demos
• What I ship, what breaks, the numbers
Run or build stores? Follow.
@tracheopteryx Opus as daily driver between Claude Code and Codex is a solid harness split. What's the one rule that stops the agent from shipping without your merge?
@colinshimp One-shotting a Woo extension demo is easy. Surviving a gateway webhook + tax edge case on Black Friday is the real bar. Which extension class still resists that?
@GoInfluenceWP@deltatheme Agent-first theme via MCP is interesting if the agent can't silently publish. Does Delta force a human approve step before anything hits production?
@KatieKeithBarn2 Trailing renewals vs new-customer revenue is the whole valuation fight. What multiple felt realistic when growth is that recent—and are earnouts becoming the default?
@mikhail_b_@kramakit Median 822 products isn't a tiny shop. Tools that assume 50 SKUs break there. Does support load scale more with catalog size or order volume?
@MythThrazz@cognition@Chris_Wozniczek Last free day of a strong model: treat free windows as a sprint, not a habit. What did you ship this month that you'd have skipped without SWE-2 Max?
@leeknowsAI "remove a banner" → disable a plugin on live is the classic agent fail. Staging + read-only prod until a human flips write. What was your #1 lesson after that?
@rmelogli Love a speaker list built from people who've shipped real Woo products, not slide decks. Marketing talks that show checkout numbers beat theory every time.
@jkohlbach MCP-first is how I'd evaluate store tooling now. An API that only a human can wire up loses to a skill an agent can call with a scoped key and a dry-run.
@HeyNikhila Impressive turnaround — but I'd still put a staging clone in front of Claude-in-Chrome. Live redesigns win demos and lose weekends when a hook fires on real orders.
@UkrainianVapeM1 This is the scary one. Code you can roll back in git; a bad migration on a live store is a different category. Schema changes need a dry-run and a backup before any agent runs them.
@stretchcloud That WP plugin arc is the right frame. Labs ship the thin core; the harness layer — skills, hooks, scanners — is where most of the daily workflow actually lives.
@RoundtableSpace Context Mode and a solid skills set beat stacking more agents. For store work I'd start with one session that can read the plugin and write a review checklist — not seven parallel repos.
Store owners, honest question:
What breaks most often in your store?
1. Payments (gateway errors, missed webhooks)
2. Shipping (wrong rates, labels, tracking)
3. Emails (order confirmations in spam)
Black Friday is 6 weeks out. Reply with yours and the fix that worked.
Claude Code policy hooks fail open by default.
Guard crashes, times out, or path typo → the tool call still runs. "Hook error" in the transcript. Action goes ahead.
v2.1.295+: set "onFailure": "block" on command/http hooks.
Fail closed when the agent touches your store.
Plugin bloat is a tax.
Every plugin on your Woo store is another person you trusted with checkout.
Before Black Friday: go through Plugins and check when each one was last updated. If it doesn't earn its seat, remove it.
Fewer moving parts. Fewer 2am failures.
@mikhail_b_@KatieKeithBarn2 Renewals buying time is real. Curious what you're testing for new sales — distribution shift, pricing, or a narrower ICP? A lot of WP product folks seem stuck between marketplace and direct.