ryanportfolio/Harness-Firmware

long-horizon

Use for work too big for one context window: long multi-step tasks, progress lost to compaction or failed retries, work spanning hours or sessions, or when the user says $long-horizon or asks to run a task in verified rounds.

原文语言：英语

verify-this

Verify a claim with fresh local evidence: restate it falsifiably, capture baseline and treatment, compare, return VERIFIED, NOT VERIFIED, or INCONCLUSIVE. Use for /verify-this, "prove it works", "did this fix it", "show me the evidence".

原文语言：英语

caveman

Ultra-compressed communication mode. Cuts token usage ~75% by speaking like caveman while keeping full technical accuracy.

原文语言：英语

humanizer

Use when the user asks to humanize, de-AI, de-slop, voice-match, or review prose for AI tells before publishing.

原文语言：英语

writing-skills

Use when creating new skills, editing existing skills, or verifying skills work before deployment

原文语言：英语

codex-review

Cross-vendor second-opinion review. Drives OpenAI Codex CLI (codex exec review, gpt-5.6-sol, high reasoning) over a PR, branch, commit, or uncommitted diff, then verifies each finding. Trigger: $codex-review, "have Codex/Sol review this".

原文语言：英语

wow-loop

Multi-agent perfection loop for any deliverable. Recon, one spec, one implementer, adversarial screenshot-verified critique until an evidence gate passes. Use when the user says $wow-loop, asks for "wow factor" or "dial it to 11".

原文语言：英语

addskill

Use whenever a repo-local skill is being added, installed, or created for future Claude Code and Codex sessions, whether the user asked or you decided to write one; writing-skills covers authoring, not this repo's install contract.

原文语言：英语

refine

Use when a session or task is wrapping up ("wrap up", "that's everything", "done for today"), when the user invokes $refine, or after any task with friction: calls wasted rediscovering a fact, a skill that misfired, a user correction.

原文语言：英语

writing-plans

Use when you have a spec or requirements for a multi-step task, before touching code

原文语言：英语

writing-skills

Use when creating new skills, editing existing skills, or verifying skills work before deployment

原文语言：英语

optimize-context

Use when the user asks to reduce per-turn context or token load, trim kernels, skills, or connectors, or propagate a generic context optimization to the starter.

原文语言：英语