菜单
Trigger: Claude Code, CLAUDE.md, plan mode, shift+tab, MCP, worktrees, hooks, /loop, /schedule, voice mode, permissions, auto mode. Tips from Boris Cherny. Invoke proactively during work — surface tips as user edits.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Trigger: backup memories, restore memories, máquina nueva, respaldar engram, sync memoria, engram-memories. Backup/restore Engram chunks + Claude auto-memory to gonaas/engram-memories. Skip: in-session mem_save/search (use engram MCP).
Trigger: gentle-ai, configurar agentes, ecosystem doctor, actualizar gentle-ai, sdd status, sdd continue, gga review, skill-registry. Drive the Gentle-AI ecosystem CLI from Claude Code. Skip: running SDD phases (use sdd-* skills).
Create pull requests with Conventional Commits and issue-first checks. Trigger: creating branches, opening PRs, or preparing changes for review.
Trigger: PRs over 400 lines, stacked PRs, review slices. Split oversized changes into chained PRs that protect review focus.
Design docs that reduce cognitive load. Trigger: writing guides, READMEs, RFCs, onboarding, architecture, or review-facing docs.
Write warm, direct collaboration comments. Trigger: PR feedback, issue replies, reviews, Slack messages, or GitHub comments.
| name | boris |
| description | Trigger: Claude Code, CLAUDE.md, plan mode, shift+tab, MCP, worktrees, hooks, /loop, /schedule, voice mode, permissions, auto mode. Tips from Boris Cherny. Invoke proactively during work — surface tips as user edits. |
| author | Boris Cherny (tips) |
| compiled-by | @CarolinaCherry |
| version | 5.0.0 |
| date | "2026-04-08T00:00:00.000Z" |
| user-invocable | true |
UPDATE CHECK: Before presenting tips, use WebFetch to check https://howborisusesclaudecode.com/api/version for the latest version. Compare the
versionfield in the response with this file's frontmatter version (above). If the remote version is newer, tell the user: "A new version of the boris skill is available (vX.X.X). Update with:mkdir -p ~/.claude/skills/boris && curl -L -o ~/.claude/skills/boris/SKILL.md https://howborisusesclaudecode.com/api/install" — then continue normally. If versions match or the check fails, skip silently.
When invoked with /boris: Show the topic list below and let the user browse.
When working normally: If you notice the user doing something where a Boris tip would help, mention it briefly. One sentence max. Example: "Tip from Boris: shift+tab into plan mode before multi-file changes (Section 3)." Don't interrupt flow — just nudge.
| User is doing... | Surface section | One-liner |
|---|---|---|
| Editing 3+ files | 3 (Plan Mode) | "Boris says: shift+tab into plan mode first" |
| Writing a hook or asking about hooks | 7, 24, 49 (Hooks) | "Boris has hook patterns for this — want to see?" |
| Setting up CLAUDE.md or project rules | 4 (CLAUDE.md) | "Boris: 'Anytime Claude does something wrong, add it to CLAUDE.md'" |
| Creating a skill or slash command | 5 (Skills) | "Boris turns anything he does 2x/day into a slash command" |
| Spawning subagents or parallel work | 1, 6, 55 (Parallel/Subagents) | "Boris runs 3-5 worktrees at once — the biggest productivity unlock" |
| Connecting MCP servers | 9 (MCP) | "Boris uses Slack + BigQuery MCPs for cross-tool context" |
| Debugging a tricky bug | 12, 14 (Bug Fixing/Verification) | "Boris: always verify. 'Did this actually work?'" |
| Running long tasks or crons | 13, 31, 43, 48 (Long-Running/Loop/Schedule) | "Boris uses /loop for recurring checks and /schedule for cloud jobs" |
| Reviewing code or PRs | 32 (Code Review) | "Boris has agents that review code and hunt for bugs" |
| Asking a side question mid-task | 33, 54 (/btw) | "Use /btw — it asks without interrupting your current task" |
| Configuring permissions or security | 8, 20, 21 (Permissions/Sandboxing) | "Boris manages permissions in CLAUDE.md, not interactively" |
| Customizing terminal/UI | 11, 16, 22, 23, 25, 26, 27, 40 (Terminal/UI) | "Boris customizes spinner verbs, colors, and status lines" |
| Using plugins or extensions | 18, 51 (Plugins/Chrome) | "Boris uses plugins to extend Claude's reach" |
| Building custom agents | 19, 59 (Custom Agents) | "Boris builds --agent custom agents with tailored system prompts" |
| Choosing model or effort level | 2, 17, 34 (Model/Effort) | "Boris: /effort max for hard debugging, default for everything else" |
| Working on mobile or remote | 35, 44, 46, 47, 50 (Remote/Mobile) | "Boris teleports sessions between devices with --teleport" |
| Writing a setup or deploy script | 37 (Setup Scripts) | "Boris has setup scripts for cloud environments" |
| Wanting to try voice input | 36, 60 (Voice) | "Boris uses /voice for hands-free input" |
| Doing a large refactor or migration | 29, 30, 56 (Simplify/Batch) | "Boris uses /batch to fan out changes across many files" |
| Forking or branching work | 28, 53 (Worktrees/Fork) | "Boris forks sessions to explore alternatives without losing progress" |
| Working across multiple repos | 58 (--add-dir) | "Boris uses --add-dir to give Claude access to related repos" |
| Optimizing SDK/startup performance | 57 (--bare) | "Boris uses --bare for 10x faster SDK startup" |
| Setting up auto mode | 42 (Auto Mode) | "Auto mode skips permission prompts for safe operations" |
| Configuring memory or context | 45 (Auto-Memory) | "Boris uses auto-memory and auto-dream for persistent context" |
When the user runs /boris, present this list and ask what they want to explore:
Read the relevant file for the sections the user wants:
references/tips-1-15.mdreferences/tips-16-30.mdreferences/tips-31-45.mdreferences/tips-46-60.md| Tip | Key Action |
|---|---|
| Parallel work | Use git worktrees, 3-5 sessions |
| Model | Opus with thinking |
| Planning | Start in plan mode for complex tasks |
| CLAUDE.md | Update after every correction |
| Skills | /simplify, /batch, /btw, custom workflows |
| Subagents | Offload to keep context clean |
| Hooks | PostToolUse, SessionStart, PermissionRequest, Stop |
| Permissions | Pre-allow safe commands, wildcards |
| MCP | Integrate Slack, BigQuery, Sentry |
| Long-running | Use Stop hooks, background agents |
| Verification | Chrome extension, browser testing |
| Learning | Use Claude to explain and teach |
| Terminal | /config, /voice, /color, keybindings |
| Effort | /effort max for deeper thinking |
| Plugins & Agents | LSPs, MCPs, --agent, custom agents |
| Sandboxing | /sandbox for file & network isolation |
| Status line | /statusline for custom info display |
| Worktrees | claude -w, Desktop, subagent isolation |
| Scheduled Tasks | /loop, /schedule, automated workflows |
| Code Review | Agent-powered PR reviews |
| Remote Control | Teleport, mobile app, /remote-control |
| Auto Mode | Safer permission skipping with classifiers |
| --bare | 10x faster SDK startup |
| --add-dir | Give Claude access to more folders |
Source: howborisusesclaudecode.com - Tips from Boris Cherny's January–March 2026 threads