WorkflowsBrew 12 · Dial-in
Get a coding agent that follows the repo's rules and stays inside its usage: write or prune the rules and guardrails, tier the models by role, then pick the effort each large task deserves.
Step 1 · pick one
Portable Harness Set up a repo so any coding agent (Claude Code, Codex, Cursor) behaves the same wherever it runs - writes 12 working rules at the top of CLAUDE.md or AGENTS.md, adds map sections only where the agent keeps getting things wrong, and… v1.0.0 Harness Audit Audits an existing coding-agent harness (CLAUDE.md, AGENTS.md, rules files, hooks, MCP and tool config, skills) against the project's past Claude Code sessions, then returns a ranked findings report plus proposed diffs applied only after… unversionedStep 2 · optional
Setup Claude Advisor Configure Claude Code for quota-optimised model tiering: Sonnet as the main session model, Opus as the on-call advisor, Haiku for read-only subagents, and Sonnet for code-editing subagents. v0.1.2Step 3 · optional
Diagnose Usage Diagnoses why a Claude Code session or day burned through usage limits faster than expected, and says what to change. v1.0.0Step 4 · optional
Choose Effort Asks the user how much effort a Claude Code task deserves before heavy work starts, then recommends an effort level, whether to fan out to subagents, goal mode, workflows or agent teams, how to set it, and when to check /usage. v1.0.0Step one is one or the other: portable-harness writes a setup, harness-audit prunes one that has grown. The usage steps are for Claude Code.
npx skills add arm-pz/portable-harness # step 1 npx skills add arm-pz/harness-audit # step 1, or the line above npx skills add arm-pz/setup-claude-advisor # step 2, optional npx skills add arm-pz/diagnose-usage # step 3, optional npx skills add arm-pz/choose-effort # step 4, optional
In Claude Code you can take every skill at once instead: /plugin marketplace add arm-pz/caffeinity-skills, then /plugin install caffeinity@caffeinity-skills. Not sure this is the right workflow? Install caffeinity-routing and describe the outcome; it picks one.
Need this fitted to your team's own tools and rules? We build custom skills.