Posts
Things I'm building, learning, seeing
Agent Skills Added a Skills Library Publisher Workflow
Added a dedicated publishing skill so future agents can keep the public skills library in sync by convention.
Agent Skills Shipped a Deploy-Blog Workflow
New skill enforces preflight checks for secrets, privacy, quality, and release gates before publishing blog content.
Agent Skills Standardized Cron Publishing Flow
Unified conventions for scheduled publishing jobs, making recurring release notes and updates repeatable.
I Built a System to Stop Re-Prompting My AI
I stopped re-prompting my AI agents and started giving them playbooks.
Default Alive Is a Strategic Weapon
The best YC advice I ever got wasn't about growth. It was about not needing anyone's money.
Don't Hire, Automate
What I'm seeing founders do instead of hiring — and the guardrail most of them skip.
Commitment, Speed, Reps
The three skills that beat abstract strategy in early-stage execution.
Falling Off the VC Track
What happens when you're profitable, capped out, and stuck in the middle.
Founder Accountability Systems That Don't Waste Everyone's Time
Most accountability groups track activity, not outcomes. Better ones center on a singular goal, weekly experiments, and decisions made.
Founder Burnout Is Not a Personal Problem
Sprinting the entire marathon isn't heroic. It's irresponsible.
The Founder Life Design Playbook
Most founders never define success. They grow for the sake of growth, grind because they're told to, and wake up with a life they didn't choose.
From 14 AI Agents to Real Pipeline
AI agents can handle your non-revenue-driving outbound work, but only after you've earned the right to automate.
Hard Lessons Building on AI Runtime Infrastructure
What happens when the thing under your thing changes every week.
Preventing AI Slop in Your Writing
Better prompts won't save you. The guardrails around the model matter more than the input to it.
Retrofitted Founder Stories
Every founder success story is a sanitized version of what actually happened. Here's what the raw version looks like.
Revenue Is Not PMF
You can grind your way to $1M ARR. That doesn't mean anyone loves your product.
The Sales Conversion of an Engineer Founder
Sales isn't pitching. It's listening. And it's the skill that unlocked everything else.
Why I Built Superclaw
What happens when you run multiple AI agents and can't see any of them.
The 52 Bets Framework
You have 52 weeks in a year. Most founders waste half of them without realizing it.
Superclaw Migrated Custom Skills to User Space
Custom skills now live in ~/.superclaw/skills/ instead of cluttering the repo, making upgrades cleaner.
Superclaw Gateway Client Never Gives Up
Gateway client now retries indefinitely instead of abandoning connection after 10 attempts.
Superclaw Now Runs as a Real Daemon
The gateway is now a persistent background service instead of a fragile dev-tied process.
Superclaw Auto-Heals Broken Tooling Config
Added self-healing for OpenClaw tools config plus a dedicated Tools settings UI.
Agent Skills Centralized Tools for Cross-Agent Portability
Tooling moved to a canonical shared surface so skills run more consistently across different agent runtimes.
Agent Skills Split Twitter Into Writer + Monitor
One Twitter skill became two focused operators: one for drafting, one for performance monitoring.
Superclaw Automation History Doesn't Disappear Anymore
Added an Automations History tab that keeps run visibility even when jobs are edited or deleted.
Superclaw Got Mobile Install + Deep Links
Added PWA support, iOS home screen install, and direct deep links into chat/task routes.
Each Superclaw Agent Now Gets Skill + API Key Provisioning
Bundled superclaw-tasks with per-agent API key auto-provisioning for safer, cleaner execution.
Agent Skills Tasks Now Execute and Return Deliverables
Tasks can now run execution commands and attach concrete outputs back into the lifecycle flow.
Agent Skills Launched Blog Writer
Added a full long-form writing skill for ideation, structure, drafting, and polish.
Agent Skills Promoted Twitter v2 Editorial Model
Twitter writing moved to an editorial-partner model with stronger idea exploration and quality constraints.
Agent Skills Shipped Scheduled Twitter Monitoring
Monitoring is now cron-driven with recurring performance briefings instead of manual checks.
Superclaw Switched to Agent Lanes for Delegation
Replaced generic status-kanban with agent lanes so delegation and ownership are obvious at a glance.
Agent Skills Made Twitter Auth Basically Hands-Off
Twitter OAuth now auto-refreshes tokens, so the workflow keeps running without repeated auth pain.