GSkillPacks
Killed: Aug 24, 2026
Cause of Death
Workflow Ceremony Overwhelmed Time to Value
Post-Mortem
GSkillPacks proved that a reusable, cross-agent skill library could be packaged, versioned, and installed through a real npm CLI, but its commercial product experience accumulated too much ceremony. Alignment pages, interrogation rounds, compiled approvals, and related workflow gates became architectural dependencies rather than optional aids, increasing the time and attention required before users reached useful work. That made the current product shape feel heavier than the problem it was meant to solve. The commercial thesis was killed without treating the underlying software as dead: the public agentic-skills repository, published skillpacks package, domain, source history, and reusable skills remain preserved as active OSS/shared infrastructure. A possible v2 would remove alignment and interrogation as required product concepts, but it remains only a revisit path until a positive low-friction experience and demand are validated.
Lessons Learned
Review and context-gathering mechanisms should earn their place through better outcomes, not become mandatory product architecture. A skill library can remain useful infrastructure even when the surrounding commercial experience is too heavy to justify continued product investment.
What I'd Do Differently
Start from immediate skill discovery, installation, and execution, then add lightweight review or clarification only when the task requires it. Do not reactivate GSkillPacks without evidence that a streamlined experience delivers faster value and solves a specific user's recurring problem better than the OSS package alone.