Stats
Actions
Tags
From team
Token/context discipline — process large data with code and surface only the answer, never read raw bulk output into context. Preload into agents that run verbose commands (tests, builds, log/data analysis). Use whenever about to read a large file, command output, or big API/tool response.
How this skill is triggered — by the user, by Claude, or both
Slash command
/team:context-disciplineThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Keep raw bulk data out of the context window. **Program the analysis — don't read everything in.** Raw output (logs, build streams, large files, full responses) is the main thing that silently fills the window and degrades quality.
Keep raw bulk data out of the context window. Program the analysis — don't read everything in. Raw output (logs, build streams, large files, full responses) is the main thing that silently fills the window and degrades quality.
grep/rg, jq, sed/awk, or a short Node/Python one-off via Bash. One script replaces many reads.cat a whole file to find one thing — grep -n -C2 (or rg) for the match with a little context.Reflex: if you're about to put more than a screenful of machine output into context, stop and script it instead.
npx claudepluginhub johantor/zion-linkCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.