Extract session discoveries and persist them as reusable skill files for Codex and Claude.
npx skills add https://github.com/sburl/CrossCheck --skill capture-skill-sburl
Please help me install this Agent Skill. Skill: capture-skill Source: https://github.com/sburl/CrossCheck/tree/main/codex/skills/capture-skill Command: npx skills add https://github.com/sburl/CrossCheck --skill capture-skill-sburl
Distill non-obvious session insights into reusable skill files that auto-load in future sessions when the trigger conditions match.
Use /capture-skill after you encounter a genuine discovery that improves workflows.
High-intent search queries and answers about installing and using this skill.
To save debugging workarounds as reusable commands, extract the non-obvious session discovery and persist it as a structured skill file. The file is stored under ~/.claude/commands/ and ~/.codex/commands/ for automatic loading in future sessions.
The best way to capture project-pattern tolerances is to distill the session insight into a structured skill file containing a unique name, trigger conditions, a step-by-step solution, and verification steps, which auto-loads when matching conditions recur.
Auto-loading captured skills works by storing the structured skill files directly in the local command caches at ~/.codex/commands/ and ~/.claude/commands/, enabling the AI to retrieve and apply the workflow automatically when trigger conditions match.
You need local access to the ~/.codex/commands/ and ~/.claude/commands/ directories to persist session discoveries. The tool enforces a structured format with trigger conditions and verification steps to ensure the skill functions correctly.
You should avoid capturing a session finding as a skill if the discovery is obvious, does not improve workflows, or represents a one-time issue that will not recur across different tasks or debugging contexts.
A captured workflow discovery must follow a structured skill format that includes a unique name, a description, specific trigger conditions, a step-by-step solution, and verification steps before being stored in the AI command caches.
ResgridRecord development discoveries into a structured learning log file.
aaioneExtract reusable patterns from Claude Code conversations and save them as learned skills.
eddiearcDelegate complex programming tasks to OpenAI Codex CLI via `codex exec --full-auto`.
chipoto69Extract non-obvious development knowledge into reusable Claude Code skills.
