Claudeception
Claudeception is a Claude Code skill that observes active coding sessions, identifies reusable patterns and successful approaches as they occur, and automatically generates new skill files capturing those patterns. The agent's institutional knowledge accumulates from its own past sessions rather than being lost at session end. The name reflects the recursive structure: Claude Code writing skills for Claude Code.
Claude Code starts each session without memory of prior sessions. Effective approaches discovered during one session — a debugging strategy, an edge case handling pattern, a refactoring sequence — are lost unless the developer manually writes a skill or updates CLAUDE.md. Claudeception automates this capture loop: it watches sessions as they run, recognises moments where a reusable pattern emerges, and writes a skill file without requiring developer intervention.
Placed in Assess at inner position because the self-improving skill extraction pattern addresses a genuine knowledge-loss problem in AI coding workflows and has no equivalent on this radar. The inner position reflects that developers who run regular Claude Code sessions should be evaluating it now — the compounding value of accumulated skills grows over time, and starting earlier yields more benefit. The pattern is distinct from static skill libraries (Anthropic Skills Repository) in that the skills are derived from the developer's own workflows rather than pre-written by others.
The quality of extracted skills depends on session quality; sessions with clear, well-scoped tasks produce more reusable patterns than exploratory or ambiguous sessions.
