RuleStack

Configs

Stacks

Compare

Diff

RuleStack

Configs

Stacks

Compare

Diff

Read API

RuleStack

Configs

Stacks

Compare

Diff

Read API

Configs/xl0/agent-files/diff

Two files, one repository

xl0/agent-files ships 2 formats across 2 indexed files. The question worth asking is whether the second one says anything the first does not.

CompareAGENTS.md ↔ CLAUDE.md
A · agents.md/AGENTS.md · 570 wordsB · agents.md/CLAUDE.md · 122 words
What each file covers, counted
DimensionSharedOnly in AOnly in BOverlap
Sections0600%
Commands0200%
Section tags0300%

What each file covers

Sections

0 shared · 6 only in A · 0 only in B
  • − State management
  • − Guidelines
  • − Tone
  • − Autonomy and persistence
  • − Code
  • − Git

Commands

0 shared · 2 only in A · 0 only in B
  • − git reset --hard
  • − git checkout --

Section tags

0 shared · 3 only in A · 0 only in B
  • − code-style
  • − git-pr
  • − do-not

Line diff

+9 added−55 removed1 unchanged1.8% identical
xl0/agent-files · agents.md/AGENTS.md
@@ −1 @@
1# State management
2 
3Maintain 2 important files in sync with the codebase:
4 
5- `CODE.md`: An in-depth summary of the current state of the codebase.
6The file should contain high-level view of the code and only non-obvious implementation details. Don't overload it with small details.
7 
8- `PLAN.md`: High-level plan in plain English, followed by TODO with [x] boxes.
9TODO items may be sections (## [x] Section) or paragraphs - don't make it rigid.
10Write down commander's intent: what needs to be done matters; how is nice to have and subject to change.
11As things are done, the plan gets compacted - paragraphs become list items, list items get merged and progressively discarded.
12 
13IMPORTANT: At the start of each conversation, always fully read `CODE.md`. You may read `PLAN.md` when relevant to the task.
14Update the files as you go, keep the updates concise. Not a changelog - content reflects the current state, not history.
15Don't put too much on one line, keep things readable.
16 
17# Guidelines
18 
19## Tone
20 
21- Be brief, be terse. Sacrifice grammar for brevity.
22- The user is very smart, knowledgeable and intelligent. Treat him like it.
23- Don't glaze the user. Correct his understanding if it's wrong. Push back on bad ideas.
24- Keep the end of turn summaries very concise.
25- No need to git diff at the end of the turn.
26 
27## Autonomy and persistence
28 
29- If the user asks for a plan, asks a question, brainstorming, or otherwise indicates conversation, reply or otherwise solve the users problem without editing the code. Otherwise, go ahead and actually implement the change. If you encounter challenges or blockers, you should attempt to resolve them yourself.
30 
31- Persist until the task is fully handled end-to-end within the current turn whenever feasible: do not stop at analysis or partial fixes; carry changes through implementation, verification, and a clear explanation of outcomes unless the user explicitly pauses or redirects you.
32 
33- If you notice unexpected changes in the worktree or staging area that you did not make, continue with your task. Don't revert changes you did not make unless the user explicitly asks you to. There can be multiple agents or the user working in the same codebase concurrently.
34 
35## Code
36 
37- Use small edits where possible. Never use sed or other hacks to edit files. Re-read and retry using tools.
38 
39- The best changes are often the smallest correct changes.
40- When you are weighing two correct approaches, prefer the more minimal one (less new names, helpers, tests, etc).
41- Keep things in one function unless composable or reusable.
42- Avoid shallow abstractions. Avoid single-use abstractions. Deep abstractions with small interface preferred.
43 
44- No speculative try/catch with fall-backs. Only handle real errors, and default to a clear explicit fail, don't implement fallbacks unless asked.
45- Never create legacy compatibility layers, unless asked specifically.
46- When experimenting or debugging, don't gate the added code - we use git, we will roll it back after experiments.
47 
48- Document data structures and interfaces, not the code.
49- Add succinct code comments that only if code is complex and not self-explanatory.
50 
51### Git
52 
53- Only commit when directly instructed.
54- When you commit, it's possible that the worktree contains unrelated changes and untracked files. Don't blindly add files - only commit what's necessary.
55- **NEVER** use destructive commands like `git reset --hard` or `git checkout --` unless specifically requested or approved by the user.
56 
xl0/agent-files · agents.md/CLAUDE.md
@@ +1 @@
1@AGENTS.md
2 
3Pi documentation (read only when the user asks about pi itself, its SDK, extensions, themes, skills, or TUI):
4- Main documentation: /home/xl0/work/projects/pi/pi-mono/packages/coding-agent/README.md
5- Additional docs: /home/xl0/work/projects/pi/pi-mono/packages/coding-agent/docs
6- Examples: /home/xl0/work/projects/pi/pi-mono/packages/coding-agent/examples (extensions, custom tools, SDK)
7- When reading pi docs or examples, resolve docs/... under Additional docs and examples/... under Examples, not the current working directory
8- When asked about: extensions (docs/extensions.md, examples/extensions/), themes (docs/themes.md), skills (docs/skills.md), prompt templates (docs/prompt-templates.md), TUI components (docs/tui.md), keybindings (docs/keybindings.md), SDK integrations (docs/sdk.md), custom providers (docs/custom-provider.md), adding models (docs/models.md), pi packages (docs/packages.md)
9- When working on pi topics, read the docs and examples, and follow .md cross-references before implementing
10- Always read pi .md files completely and follow links to related docs (e.g., tui.md for TUI API details)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
@@ −1 +1 @@
1−# State management
1+@AGENTS.md
22  
3−Maintain 2 important files in sync with the codebase:
4− 
5−- `CODE.md`: An in-depth summary of the current state of the codebase.
6−The file should contain high-level view of the code and only non-obvious implementation details. Don't overload it with small details.
7− 
8−- `PLAN.md`: High-level plan in plain English, followed by TODO with [x] boxes.
9−TODO items may be sections (## [x] Section) or paragraphs - don't make it rigid.
10−Write down commander's intent: what needs to be done matters; how is nice to have and subject to change.
11−As things are done, the plan gets compacted - paragraphs become list items, list items get merged and progressively discarded.
12− 
13−IMPORTANT: At the start of each conversation, always fully read `CODE.md`. You may read `PLAN.md` when relevant to the task.
14−Update the files as you go, keep the updates concise. Not a changelog - content reflects the current state, not history.
15−Don't put too much on one line, keep things readable.
16− 
17−# Guidelines
18− 
19−## Tone
20− 
21−- Be brief, be terse. Sacrifice grammar for brevity.
22−- The user is very smart, knowledgeable and intelligent. Treat him like it.
23−- Don't glaze the user. Correct his understanding if it's wrong. Push back on bad ideas.
24−- Keep the end of turn summaries very concise.
25−- No need to git diff at the end of the turn.
26− 
27−## Autonomy and persistence
28− 
29−- If the user asks for a plan, asks a question, brainstorming, or otherwise indicates conversation, reply or otherwise solve the users problem without editing the code. Otherwise, go ahead and actually implement the change. If you encounter challenges or blockers, you should attempt to resolve them yourself.
30− 
31−- Persist until the task is fully handled end-to-end within the current turn whenever feasible: do not stop at analysis or partial fixes; carry changes through implementation, verification, and a clear explanation of outcomes unless the user explicitly pauses or redirects you.
32− 
33−- If you notice unexpected changes in the worktree or staging area that you did not make, continue with your task. Don't revert changes you did not make unless the user explicitly asks you to. There can be multiple agents or the user working in the same codebase concurrently.
34− 
35−## Code
36− 
37−- Use small edits where possible. Never use sed or other hacks to edit files. Re-read and retry using tools.
38− 
39−- The best changes are often the smallest correct changes.
40−- When you are weighing two correct approaches, prefer the more minimal one (less new names, helpers, tests, etc).
41−- Keep things in one function unless composable or reusable.
42−- Avoid shallow abstractions. Avoid single-use abstractions. Deep abstractions with small interface preferred.
43− 
44−- No speculative try/catch with fall-backs. Only handle real errors, and default to a clear explicit fail, don't implement fallbacks unless asked.
45−- Never create legacy compatibility layers, unless asked specifically.
46−- When experimenting or debugging, don't gate the added code - we use git, we will roll it back after experiments.
47− 
48−- Document data structures and interfaces, not the code.
49−- Add succinct code comments that only if code is complex and not self-explanatory.
50− 
51−### Git
52− 
53−- Only commit when directly instructed.
54−- When you commit, it's possible that the worktree contains unrelated changes and untracked files. Don't blindly add files - only commit what's necessary.
55−- **NEVER** use destructive commands like `git reset --hard` or `git checkout --` unless specifically requested or approved by the user.
56− 
3+Pi documentation (read only when the user asks about pi itself, its SDK, extensions, themes, skills, or TUI):
4+- Main documentation: /home/xl0/work/projects/pi/pi-mono/packages/coding-agent/README.md
5+- Additional docs: /home/xl0/work/projects/pi/pi-mono/packages/coding-agent/docs
6+- Examples: /home/xl0/work/projects/pi/pi-mono/packages/coding-agent/examples (extensions, custom tools, SDK)
7+- When reading pi docs or examples, resolve docs/... under Additional docs and examples/... under Examples, not the current working directory
8+- When asked about: extensions (docs/extensions.md, examples/extensions/), themes (docs/themes.md), skills (docs/skills.md), prompt templates (docs/prompt-templates.md), TUI components (docs/tui.md), keybindings (docs/keybindings.md), SDK integrations (docs/sdk.md), custom providers (docs/custom-provider.md), adding models (docs/models.md), pi packages (docs/packages.md)
9+- When working on pi topics, read the docs and examples, and follow .md cross-references before implementing
10+- Always read pi .md files completely and follow links to related docs (e.g., tui.md for TUI API details)

Also from Kynth Studios

Built for the same person as RuleStack

ToolDrift

What the AI coding tools changed last night

tooldrift.kynth.studio

StillShipping

Which agent tools have stopped shipping

stillshipping.kynth.studio

BlockDex

Search inside every shadcn registry

blockdex.kynth.studio

The studio list

One product, taken apart, once a month

Kynth Studios pulls one shipped product open every month — what it does, what it cost to build, what the pipeline behind it looks like, and what the numbers did. One email a month, nothing in between.

Double opt-in — we send one confirmation link and nothing else until you click it.

RuleStack

Built by

Kynth Studios

the studio behind ToolDrift, StillShipping and BlockDex

part of Toolproof, the measurement layer for AI agent tooling

Directory

Configs
Stacks
Compare formats
AGENTS.md vs CLAUDE.md
Cursor rules alternatives
Diff two configs
Best AGENTS.md examples
Best Cursor rules examples
What goes in a CLAUDE.md

Formats

AGENTS.md
CLAUDE.md
Cursor rules
Copilot instructions

Reference

Read API
Corpus health
Privacy Policy
Terms

© 2026 RuleStack. A Kynth Studios product. Changelog

RuleStack

The studio list

One product, taken apart, once a month

Kynth Studios pulls one shipped product open every month — what it does, what it cost to build, what the pipeline behind it looks like, and what the numbers did. One email a month, nothing in between.

Double opt-in — we send one confirmation link and nothing else until you click it.

RuleStack

Built by

Kynth Studios

the studio behind ToolDrift, StillShipping and BlockDex

part of Toolproof, the measurement layer for AI agent tooling

Directory

Configs
Stacks
Compare formats
AGENTS.md vs CLAUDE.md
Cursor rules alternatives
Diff two configs
Best AGENTS.md examples
Best Cursor rules examples
What goes in a CLAUDE.md

Formats

AGENTS.md
CLAUDE.md
Cursor rules
Copilot instructions

Reference

Read API
Corpus health
Privacy Policy
Terms

© 2026 RuleStack. A Kynth Studios product. Changelog

RuleStack

The studio list

One product, taken apart, once a month

Kynth Studios pulls one shipped product open every month — what it does, what it cost to build, what the pipeline behind it looks like, and what the numbers did. One email a month, nothing in between.

Double opt-in — we send one confirmation link and nothing else until you click it.

RuleStack

Built by

Kynth Studios

the studio behind ToolDrift, StillShipping and BlockDex

part of Toolproof, the measurement layer for AI agent tooling

Directory

Configs
Stacks
Compare formats
AGENTS.md vs CLAUDE.md
Cursor rules alternatives
Diff two configs
Best AGENTS.md examples
Best Cursor rules examples
What goes in a CLAUDE.md

Formats

AGENTS.md
CLAUDE.md
Cursor rules
Copilot instructions

Reference

Read API
Corpus health
Privacy Policy
Terms

© 2026 RuleStack. A Kynth Studios product. Changelog

RuleStack