Evidence-first process skills for coding agents: codebase analysis, solution design, minimal implementation plans
Skill Claude CodeCodex
Use when explicitly asked for $codebase-analysis to create or fully update CURRENTSTATE.md with evidence for the current repository or focused area. Do not use for casual code questions, reviews, implementation, debugging, solution design, or planning.
Skill Claude CodeCodex
Use when explicitly asked for $debug, or when a bug survived the first fix, is intermittent, or is a performance regression: build a feedback loop that goes red on the bug, minimise, rank falsifiable hypotheses, instrument, fix with a regression test at a correct seam, clean up. Do not use for a first-read error that…
Skill Claude CodeCodex
Use when explicitly asked for $ops-review to review an implemented change for silent operational failures: missing timeouts, unbounded resources, connection leaks, non-idempotent retries, slow-dependency degradation. Read-only findings report; does not fix code and does not replace load testing.
Skill Claude CodeCodex
Use when explicitly asked for $planf3 to create, update, or build a minimal Markdown implementation plan from SOLUTION.md or a concrete approved task. Do not use for solution design or speculative architecture.
Skill Claude CodeCodex
Use when explicitly asked for $scope-review to review a diff for excess scope before a PR: additions without a forcing requirement, foreign patterns, indirection layers, duplication, files outside the task frame. Read-only findings report; does not fix code.
Skill Claude CodeCodex
Use when explicitly asked for $solution-design to create SOLUTION.md for one concrete software change before implementation planning. Produces a minimal evidence-based design; does not write code or create an implementation plan.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: