You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+57-22Lines changed: 57 additions & 22 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -44,38 +44,45 @@ This curated collection fills in those gaps. It's a living document, and you're
44
44
45
45
## Requirements & Planning
46
46
47
-
### Read, Plan, Code, Commit
48
-
49
-
Make it explore the code, then make a plan, implement it, and commit.
50
-
51
-
> "There's a process that I call 'priming' the agent, where instead of having the agent jump straight to performing a task, I have it read additional context upfront to increase the chances that it will produce good outputs."
Create context files that persistently guide tools about your project's structure, standards, and preferences.
57
50
58
51
> "`CLAUDE.md` is a special file that Claude automatically pulls into context when starting a conversation. This makes it an ideal place for documenting: common bash commands, core files and utility functions, code style guidelines, testing instructions."
**Claude Code**: Run `/init` to auto-generate smart context files that make Claude understand your codebase instantly. Claude analyzes dependencies, scripts, and architecture automatically. Use `/memory` to edit and add project-specific gotchas.
59
+
Run `/init` to auto-generate smart context files that make Claude understand your codebase instantly. Claude analyzes dependencies, scripts, and architecture automatically. Use `/memory` to edit and add project-specific gotchas.
65
60
66
61
- For new projects: Run `/init` in your project root to create a starter `CLAUDE.md`
67
62
- For existing codebases: Run `/init` and Claude will analyze your project structure, dependencies, and configuration files to automatically generate essential information
68
63
- Use `/memory` for full editor interface or `#` as quick shortcut to add notes
69
64
- Memory file hierarchy: `~/.claude/CLAUDE.md` (global) and `./CLAUDE.md` (project-specific)
70
65
71
-
**Cursor**: Create `AGENTS.md` for project rules, then use @codebase and @docs for dynamic context. Cursor's superpower is real-time understanding of your entire codebase through @-mentions.
66
+
</details>
67
+
68
+
<details>
69
+
<summary><strong>Cursor</strong></summary>
70
+
71
+
Create `AGENTS.md` for project rules, then use @codebase and @docs for dynamic context. Cursor's superpower is real-time understanding of your entire codebase through @-mentions.
72
72
73
73
- Create `AGENTS.md` at project root (also supports legacy `.cursorrules`)
74
74
- Real-time context: Use `@codebase` to pull in relevant files automatically, `@docs` to reference documentation, `@git` to understand recent changes
75
75
- Combine static rules (AGENTS.md) with dynamic context (@-mentions) for best results
76
76
77
77
</details>
78
78
79
+
### Read, Plan, Code, Commit
80
+
81
+
Make it explore the code, then make a plan, implement it, and commit.
82
+
83
+
> "There's a process that I call 'priming' the agent, where instead of having the agent jump straight to performing a task, I have it read additional context upfront to increase the chances that it will produce good outputs."
Give comprehensive specs - even a conversational spec beats vague instructions.
@@ -276,18 +283,25 @@ Stop waiting for one AI agent to finish before starting another - run multiple a
276
283
> "I disable all permission checks. Which basically means I run claude --dangerously-skip-permissions. More specifically I have an alias called claude-yolo set up."
**Claude Code**: Use `--dangerously-skip-permissions` on launch or `/permissions` to change permission strategy during coding session.
291
+
Use `--dangerously-skip-permissions` on launch or `/permissions` to change permission strategy during coding session.
283
292
284
293
- Run `claude --dangerously-skip-permissions` to enable autonomous mode where Claude runs uninterrupted without permission prompts
285
294
- Switch during session: Use `/permissions` to manage tool permissions mid-session without restarting
286
295
- When to use: Fixing lint errors across multiple files, simple refactoring and variable renames, routine code updates and migrations
287
296
- Safety: Best used in containers or VMs for isolation, avoid on critical production systems
288
297
- Setup alias: Many users create `alias cc='claude --dangerously-skip-permissions'` for quick access
289
298
290
-
**Codex CLI**: Use `--full-auto` on launch or `/mode` to change permission strategy during coding session.
299
+
</details>
300
+
301
+
<details>
302
+
<summary><strong>Codex CLI</strong></summary>
303
+
304
+
Use `--full-auto` on launch or `/mode` to change permission strategy during coding session.
291
305
292
306
- Enable full autonomous mode with `codex --full-auto` or use in-session `/mode` command
293
307
- Switch during session: Use `/mode` to hot-swap between permission levels without losing session context
@@ -310,16 +324,23 @@ Begin with faster/cheaper models for routine tasks, then escalate to more powerf
310
324
> "Sonnet 4 handles 90% of tasks effectively. Switch to Opus when Sonnet gets stuck. Recommend starting with Sonnet and providing comprehensive context."
**Claude Code**: Use `/model` to switch between models based on task complexity. Start with cheaper Sonnet 4 for routine work, escalate to fast Opus 4.1 when stuck.
332
+
Use `/model` to switch between models based on task complexity. Start with cheaper Sonnet 4 for routine work, escalate to fast Opus 4.1 when stuck.
317
333
318
334
- Use `/model` to switch models during your session
319
335
- Cheaper, faster option: `Claude Sonnet 4`
320
336
- Top-graded option: `Claude Opus 4.1`
321
337
322
-
**Codex CLI**: Use `/model` to escalate when needed. Start with `gpt-5-medium` for most tasks, switch to `gpt-5-high` when you hit complex problems.
338
+
</details>
339
+
340
+
<details>
341
+
<summary><strong>Codex CLI</strong></summary>
342
+
343
+
Use `/model` to escalate when needed. Start with `gpt-5-medium` for most tasks, switch to `gpt-5-high` when you hit complex problems.
323
344
324
345
- Use `/model` to switch models during your session
325
346
- Cheaper, faster option: `gpt-5-medium`
@@ -341,16 +362,23 @@ Reset the AI's context window between unrelated tasks to prevent confusion and i
341
362
> "During long sessions, Claude's context window can fill with irrelevant conversation, file contents, and commands. This can reduce performance and sometimes distract Claude. Use the `/clear` command frequently between tasks to reset the context window."
**Claude Code**: Use `/clear` command to reset the context window and start fresh.
370
+
Use `/clear` command to reset the context window and start fresh.
348
371
349
372
- Run `/clear` between unrelated tasks to prevent context pollution
350
373
- Preserves your memory files (CLAUDE.md) while clearing conversation history
351
374
- Essential for long coding sessions to maintain performance
352
375
353
-
**Cursor**: Start new chat sessions or use conversation management features.
376
+
</details>
377
+
378
+
<details>
379
+
<summary><strong>Cursor</strong></summary>
380
+
381
+
Start new chat sessions or use conversation management features.
354
382
355
383
- Use Cmd/Ctrl+Shift+L to start a new chat for unrelated tasks
356
384
- Chat history is automatically managed to prevent context overload
@@ -372,17 +400,24 @@ Don't let AI go too far down the wrong path - interrupt, provide feedback, and r
372
400
> "Press Escape to interrupt Claude during any phase (thinking, tool calls, file edits), preserving context so you can redirect or expand instructions. Double-tap Escape to jump back in history, edit a previous prompt, and explore a different direction. You can edit the prompt and repeat until you get the result you're looking for."
0 commit comments