Claude Code Tips

> [EN](./README.md) | [ZH](./README.zh-CN.md) | [ES](./README.es.md) | [HI](./README.hi.md) | [PT](./README.pt-BR.md) | [JA](./README.ja.md)

claude-code-tips

[](https://github.com/anipotts/claude-code-tips/actions/workflows/validate.yml) [](https://github.com/anipotts/claude-code-tips/stargazers) [](https://github.com/anipotts/claude-code-tips/commits/main) [](https://docs.anthropic.com/en/docs/claude-code) [](./LICENSE)

claude code patterns, battle-tested across yc startups, public tech companies, and unicorns. maintained by someone who uses claude code as their job.

new here? start with the [tips index](./docs/tips/) or skim [hooks](./docs/hooks.md) and [automation](./docs/automation.md).

what's inside

three plugins, one marketplace.

• lore@anipotts every session mined into sqlite. query costs, tools, errors, hotspots, loops, and full-text search across your own history. everything local.
• cc@anipotts cross-session awareness and messaging. plus a time subsystem: /cc:time-estimate gives realistic claude-code time grounded in your session history, not optimistic guesses.
• time@anipotts 3-meter fuel gauge (5-hour session, 7-day weekly, 200k context). pre-turn hook nudges claude toward cleaner handoffs as the meters fill. /fuel state reads them directly; /fuel handoff drafts a stopping point.

> /cc:time-estimate "rewrite auth middleware and add tests"
CC: ~22 min active (standard mode, Opus 4.7 high)
your time: ~15 min review

quick start

/plugin marketplace add anipotts/claude-code-tips   # add marketplace (one time)
/plugin install lore@anipotts                             # install lore (session analytics)
/plugin install cc@anipotts                               # install cc (cross-session messaging)
/plugin install time@anipotts                             # install time (fuel gauge + estimates)

then: copy [safety-guard.sh](./hooks/safety-guard.sh) to block dangerous commands. read a [tip](./docs/tips/). done.

---

the numbers

hundreds of sessions across dozens of projects. $200/mo max plan.

same usage would cost ~$12K on the API with caching, ~$95K without. no autonomous loops. no cron jobs. every session starts with me typing a prompt. [how the cost math works →](./docs/cost.md)

---

install the lore plugin

/plugin marketplace add anipotts/claude-code-tips   # add marketplace (one time)
/plugin install lore@anipotts                             # install lore (session analytics)
/plugin install cc@anipotts                               # install cc (cross-session messaging)
/plugin install time@anipotts                             # install time (fuel gauge + estimates)

you get [lore](./plugins/lore/) · session mining to sqlite. costs, search, error memory, pattern detection. all data stays local at ~/.claude/lore/lore.db.

/lore                     today's sessions, cost, top tools
/lore search "websocket"  full-text search across all conversations
/lore mistakes            error patterns claude keeps repeating
/lore hotspots            most-edited files across sessions
/lore loops               repeated patterns across sessions

start with lore + the safety-guard hook. add more as you go. [lore docs →](./plugins/lore/)

---

cc plugin

cross-session messaging and the time subsystem. see what other claude code sessions are doing, send messages between them, and get realistic time estimates grounded in your own session history.

/plugin install cc@anipotts

/cc                             show active sessions
/cc send merizo "pause"         message another session
/cc:time-estimate         ranged CC estimate, uses your current model + effort
/cc:time-calibrate              diff real throughput (from lore.db) against the rule
/cc:time-benchmark              guided A/B/C across effort levels on your model

---

the 3 things that changed how i code

hooks

hooks are the difference between "claude does what i want" and "claude does whatever it feels like." CLAUDE.md gives guidance. hooks give enforcement. one is a suggestion, the other is a wall.

this repo has 9 hooks you can drop into any project. safety-guard blocks force pushes, rm -rf /, and curl | bash. no-squash blocks squash merges. context-save preserves state before compaction. pick the ones that fit your workflow. [hook guide →](./docs/hooks.md)

agent teams

multiple claude instances working simultaneously on the same codebase, each in its own git worktree. the coordinator assigns tasks, collects results, merges the best approach.

i use this for parallel research, trying risky changes safely, and comparing approaches side-by-side without touching my working tree. [how i use agent teams →](./docs/agents.md)

prompt caching

this is why the $200/mo plan is the best deal in AI coding. claude code caches your system prompt, tools, and CLAUDE.md as a prefix. 91% of my input tokens hit the cache, meaning i pay 10% of the input cost on 91% of my reads.

the key: keep your CLAUDE.md short and stable. every edit breaks the prefix cache. mine is 30 lines and changes maybe once a week. [the full cost breakdown →](./docs/cost.md)

---

tips

short, standalone techniques. each one is something you can use in your next session.

| tip | what you learn | |-----|---------------| | [prompt caching](./docs/tips/prompt-caching.md) | get 97%+ cache hit rates, slash your bill | | [safety hooks](./docs/tips/safety-hooks.md) | block force pushes and rm -rf in 5 minutes | | [settings hierarchy](./docs/tips/settings-hierarchy.md) | project vs global vs local settings | | [session length](./docs/tips/session-length.md) | why shorter sessions are more efficient (with data) | | [ultrathink](./docs/tips/ultrathink.md) | force extended thinking for complex problems | | [context management](./docs/tips/context-management.md) | compaction strategies, active tool rate, keeping sessions tight | | [plan mode](./docs/tips/plan-mode.md) | when planning saves time vs when it wastes it | | [fast mode](./docs/tips/fast-mode.md) | same model, faster output, the tradeoff | | [plugins](./docs/tips/plugins.md) | build a plugin from scratch, what makes one worth installing | | [subagents](./docs/tips/subagents.md) | agent teams, worktree isolation, when parallel pays off | | [mcp integration](./docs/tips/mcp-integration.md) | wire up MCP servers, use them inside sessions | | [hooks v2](./docs/tips/hooks-v2.md) | command vs http vs prompt hooks, the async pattern |

---

hooks

copy one, wire it up, done. each is a standalone bash script. [full guide →](./docs/hooks.md)

| hook | event | what it does | |---|---|---| | [safety-guard](./hooks/safety-guard.sh) | PreToolUse | blocks force push, rm -rf /, DROP TABLE, curl-pipe-sh | | [no-squash](./hooks/no-squash.sh) | PreToolUse | blocks squash merges | | [panopticon](./hooks/panopticon.sh) | PostToolUse | logs every tool call to sqlite | | [context-save](./hooks/context-save.sh) | PreCompact | saves context before compression | | [notify](./hooks/notify.sh) | Notification | routes to macOS, Slack, ntfy |

4 more hooks

| hook | event | what it does | |---|---|---| | [commit-nudge](./hooks/commit-nudge.sh) | PostToolUse | reminds you to commit after N edits | | [version-stamp](./hooks/version-stamp.sh) | SessionEnd | auto-updates "tested with" stamps | | [stale-branch](./hooks/stale-branch.sh) | SessionStart | warns about gone tracking branches | | [md-lint-fix](./hooks/md-lint-fix.sh) | PostToolUse | auto-fixes markdown lint on save |

example agents

copy to .claude/agents/ and invoke with /agent . each teaches a different pattern. [guide →](./docs/agents.md)

| agent | pattern | what it does | |---|---|---| | [watch-tests](./examples/agents/watch-tests.md) | daemon | watches files, runs tests, proposes fixes | | [try-worktree](./examples/agents/try-worktree.md) | worktree | tries risky changes in isolated worktrees | | [arch-review](./examples/agents/arch-review.md) | quick review | fast architecture smell-test | | [write-pr](./examples/agents/write-pr.md) | git integration | PR descriptions from your diff |

commands i use

| command | what it does | |---|---| | /lore | usage data · costs, sessions, search, patterns | | /ship | stage, commit, push, open PR in one command | | /improve | propose CLAUDE.md updates from git history |

plus [2 example commands](./examples/commands/) you can copy: /sweep, /quicktest.

---

my personal takes

| | what | |---|---| | [cost reality](./docs/cost.md) | what claude code actually costs, the prompt caching math | | [mistakes i made](./docs/mistakes.md) | what burned me so you can skip it | | [automation](./docs/automation.md) | patterns for safe daemons, cron, and github actions with claude code | | [session workflow](./docs/session-workflow.md) | how i work day-to-day with claude code | | [worktrees](./docs/worktrees.md) | parallel exploration with the desktop app |

vs the alternatives

diplomatic, data-driven, no FUD. every claim cites a source.

[vs cursor](./docs/comparisons/cursor.md) · [vs codex](./docs/comparisons/codex.md) · [vs gemini](./docs/comparisons/gemini.md) · [vs antigravity](./docs/comparisons/antigravity.md) · [pricing](./docs/comparisons/pricing.md)

---

examples

• [CLAUDE.md templates](./examples/claude-md/) · starter configs for TypeScript, Python, Rust, Next.js
• [example agents](./examples/agents/) · 4 agents, each teaching a different pattern
• [example commands](./examples/commands/) · 2 commands you can copy to any project
• [handoff plugin](./examples/plugins/handoff/) · PreCompact context preservation
• [broadcast plugin](./examples/plugins/broadcast/) · async notifications on git events

---

how this repo works

this repo runs on its own patterns.

• CI workflows for validation, freshness, competitive intel, community signal, releases, AI-assisted review
• 9 standalone hooks + per-plugin hooks across cc / lore / time
• AI-powered checks use haiku so monthly CI cost stays under a couple of dollars
• automation handles the mechanical work so judgment-heavy decisions stay human

[automation philosophy →](./docs/automation.md)

---

tools i built from these patterns

these all came out of living in claude code every day. each solves a specific problem i kept hitting.

• [lore](./plugins/lore/) · session mining to sqlite. costs, search, error memory, pattern detection
• claudemon · real-time session monitoring across projects and machines
• [cc](./plugins/cc/) · multi-session awareness. see what other sessions are doing, send messages between them
• imessage-mcp · MCP server for read-only iMessage history. 26 tools, zero network requests

more from me

• anipotts.com/thoughts · long-form
• buttondown.com/anipotts · newsletter
• @anipottsbuilds · short-form

---

MIT · built by anipotts

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: anipotts
• Source: anipotts/claude-code-tips
• License: MIT
• Homepage: https://anipotts.com

Install and usage instructions live in the source repository linked above.