meta: add minimal AGENTS.md

[BethGriggs]

Now that the project’s AI use guidelines have landed (#62105), which permit the use of such tools, this proposes adding an AGENTS.md file to direct coding agents to the existing policies before they make changes.

Some tools may already discover and follow these documents, but this should make that behaviour slightly more deterministic. If it proves useful then we could consider extending with further instructions.

One consideration is that AGENTS.md is currently ignored (#62612). So, some contributors may have their own file which would subsequently conflict.

I know opinions vary, but the intent here is not to encourage usage of AI tools further than #62105, but to take the opportunity to guide the output.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.13%. Comparing base (21f0f27) to head (9e5a6ae).
⚠️ Report is 12 commits behind head on main.

Additional details and impacted files

@@ Coverage Diff @@## main #65463 +/- ##
==========================================
+ Coverage 90.12% 90.13% +0.01% 
==========================================
Files 752 751 -1 Lines 252315 252298 -17 Branches 47444 47437 -7 ==========================================
+ Hits 227395 227411 +16 + Misses 16217 16200 -17 + Partials 8703 8687 -16 

see 32 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[bmuenzenmeyer]

I'm tempted to adopt the convention of suggesting a CLAUDE.md that only states @AGENTS.md per many community workarounds - but I like more that we support the open standard only instead of supporting proprietary formats.

[MikeMcC399]

#65493 from @jasnell is taking a different approach, adding an AGENTS.md file with much more than just minimal contents. This would probably be preferable and it appears there is growing consensus on the content.

[jasnell]

I actually hadn't noticed this when I opened my alternative. I went with the more instructive version largely because I've noticed that some agents are... well, less inteligent than others when it comes to proactively following links and instructions unless they are spelled out in specifics. Either way, let's get something in.

[BethGriggs]

Ah, probably superseded by #65493

[BethGriggs]

Now that the project’s AI use guidelines have landed (#62105), which permit the use of such tools, this proposes adding an AGENTS.md file to direct coding agents to the existing policies before they make changes.

Some tools may already discover and follow these documents, but this should make that behaviour slightly more deterministic. If it proves useful then we could consider extending with further instructions.

One consideration is that AGENTS.md is currently ignored (#62612). So, some contributors may have their own file which would subsequently conflict.

I know opinions vary, but the intent here is not to encourage usage of AI tools further than #62105, but to take the opportunity to guide the output.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.13%. Comparing base (21f0f27) to head (9e5a6ae).
⚠️ Report is 12 commits behind head on main.

Additional details and impacted files

@@ Coverage Diff @@## main #65463 +/- ##
==========================================
+ Coverage 90.12% 90.13% +0.01% 
==========================================
Files 752 751 -1 Lines 252315 252298 -17 Branches 47444 47437 -7 ==========================================
+ Hits 227395 227411 +16 + Misses 16217 16200 -17 + Partials 8703 8687 -16 

see 32 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[bmuenzenmeyer]

I'm tempted to adopt the convention of suggesting a CLAUDE.md that only states @AGENTS.md per many community workarounds - but I like more that we support the open standard only instead of supporting proprietary formats.

[MikeMcC399]

#65493 from @jasnell is taking a different approach, adding an AGENTS.md file with much more than just minimal contents. This would probably be preferable and it appears there is growing consensus on the content.

[jasnell]

I actually hadn't noticed this when I opened my alternative. I went with the more instructive version largely because I've noticed that some agents are... well, less inteligent than others when it comes to proactively following links and instructions unless they are spelled out in specifics. Either way, let's get something in.

[BethGriggs]

Ah, probably superseded by #65493

[BethGriggs]

Now that the project’s AI use guidelines have landed (#62105), which permit the use of such tools, this proposes adding an AGENTS.md file to direct coding agents to the existing policies before they make changes.

Some tools may already discover and follow these documents, but this should make that behaviour slightly more deterministic. If it proves useful then we could consider extending with further instructions.

One consideration is that AGENTS.md is currently ignored (#62612). So, some contributors may have their own file which would subsequently conflict.

I know opinions vary, but the intent here is not to encourage usage of AI tools further than #62105, but to take the opportunity to guide the output.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.13%. Comparing base (21f0f27) to head (9e5a6ae).
⚠️ Report is 12 commits behind head on main.

Additional details and impacted files

@@ Coverage Diff @@## main #65463 +/- ##
==========================================
+ Coverage 90.12% 90.13% +0.01% 
==========================================
Files 752 751 -1 Lines 252315 252298 -17 Branches 47444 47437 -7 ==========================================
+ Hits 227395 227411 +16 + Misses 16217 16200 -17 + Partials 8703 8687 -16 

see 32 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[bmuenzenmeyer]

I'm tempted to adopt the convention of suggesting a CLAUDE.md that only states @AGENTS.md per many community workarounds - but I like more that we support the open standard only instead of supporting proprietary formats.

[MikeMcC399]

#65493 from @jasnell is taking a different approach, adding an AGENTS.md file with much more than just minimal contents. This would probably be preferable and it appears there is growing consensus on the content.

[jasnell]

I actually hadn't noticed this when I opened my alternative. I went with the more instructive version largely because I've noticed that some agents are... well, less inteligent than others when it comes to proactively following links and instructions unless they are spelled out in specifics. Either way, let's get something in.

[BethGriggs]

Ah, probably superseded by #65493

[BethGriggs]

Now that the project’s AI use guidelines have landed (#62105), which permit the use of such tools, this proposes adding an AGENTS.md file to direct coding agents to the existing policies before they make changes.

Some tools may already discover and follow these documents, but this should make that behaviour slightly more deterministic. If it proves useful then we could consider extending with further instructions.

One consideration is that AGENTS.md is currently ignored (#62612). So, some contributors may have their own file which would subsequently conflict.

I know opinions vary, but the intent here is not to encourage usage of AI tools further than #62105, but to take the opportunity to guide the output.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.13%. Comparing base (21f0f27) to head (9e5a6ae).
⚠️ Report is 12 commits behind head on main.

Additional details and impacted files

@@ Coverage Diff @@## main #65463 +/- ##
==========================================
+ Coverage 90.12% 90.13% +0.01% 
==========================================
Files 752 751 -1 Lines 252315 252298 -17 Branches 47444 47437 -7 ==========================================
+ Hits 227395 227411 +16 + Misses 16217 16200 -17 + Partials 8703 8687 -16 

see 32 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[bmuenzenmeyer]

I'm tempted to adopt the convention of suggesting a CLAUDE.md that only states @AGENTS.md per many community workarounds - but I like more that we support the open standard only instead of supporting proprietary formats.

[MikeMcC399]

#65493 from @jasnell is taking a different approach, adding an AGENTS.md file with much more than just minimal contents. This would probably be preferable and it appears there is growing consensus on the content.

[jasnell]

I actually hadn't noticed this when I opened my alternative. I went with the more instructive version largely because I've noticed that some agents are... well, less inteligent than others when it comes to proactively following links and instructions unless they are spelled out in specifics. Either way, let's get something in.

[BethGriggs]

Ah, probably superseded by #65493

[BethGriggs]

Now that the project’s AI use guidelines have landed (#62105), which permit the use of such tools, this proposes adding an AGENTS.md file to direct coding agents to the existing policies before they make changes.

Some tools may already discover and follow these documents, but this should make that behaviour slightly more deterministic. If it proves useful then we could consider extending with further instructions.

One consideration is that AGENTS.md is currently ignored (#62612). So, some contributors may have their own file which would subsequently conflict.

I know opinions vary, but the intent here is not to encourage usage of AI tools further than #62105, but to take the opportunity to guide the output.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.13%. Comparing base (21f0f27) to head (9e5a6ae).
⚠️ Report is 12 commits behind head on main.

Additional details and impacted files

@@ Coverage Diff @@## main #65463 +/- ##
==========================================
+ Coverage 90.12% 90.13% +0.01% 
==========================================
Files 752 751 -1 Lines 252315 252298 -17 Branches 47444 47437 -7 ==========================================
+ Hits 227395 227411 +16 + Misses 16217 16200 -17 + Partials 8703 8687 -16 

see 32 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[bmuenzenmeyer]

I'm tempted to adopt the convention of suggesting a CLAUDE.md that only states @AGENTS.md per many community workarounds - but I like more that we support the open standard only instead of supporting proprietary formats.

[MikeMcC399]

#65493 from @jasnell is taking a different approach, adding an AGENTS.md file with much more than just minimal contents. This would probably be preferable and it appears there is growing consensus on the content.

[jasnell]

I actually hadn't noticed this when I opened my alternative. I went with the more instructive version largely because I've noticed that some agents are... well, less inteligent than others when it comes to proactively following links and instructions unless they are spelled out in specifics. Either way, let's get something in.

[BethGriggs]

Ah, probably superseded by #65493

[BethGriggs]

Now that the project’s AI use guidelines have landed (#62105), which permit the use of such tools, this proposes adding an AGENTS.md file to direct coding agents to the existing policies before they make changes.

Some tools may already discover and follow these documents, but this should make that behaviour slightly more deterministic. If it proves useful then we could consider extending with further instructions.

One consideration is that AGENTS.md is currently ignored (#62612). So, some contributors may have their own file which would subsequently conflict.

I know opinions vary, but the intent here is not to encourage usage of AI tools further than #62105, but to take the opportunity to guide the output.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.13%. Comparing base (21f0f27) to head (9e5a6ae).
⚠️ Report is 12 commits behind head on main.

Additional details and impacted files

@@ Coverage Diff @@## main #65463 +/- ##
==========================================
+ Coverage 90.12% 90.13% +0.01% 
==========================================
Files 752 751 -1 Lines 252315 252298 -17 Branches 47444 47437 -7 ==========================================
+ Hits 227395 227411 +16 + Misses 16217 16200 -17 + Partials 8703 8687 -16 

see 32 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[bmuenzenmeyer]

I'm tempted to adopt the convention of suggesting a CLAUDE.md that only states @AGENTS.md per many community workarounds - but I like more that we support the open standard only instead of supporting proprietary formats.

[MikeMcC399]

#65493 from @jasnell is taking a different approach, adding an AGENTS.md file with much more than just minimal contents. This would probably be preferable and it appears there is growing consensus on the content.

[jasnell]

I actually hadn't noticed this when I opened my alternative. I went with the more instructive version largely because I've noticed that some agents are... well, less inteligent than others when it comes to proactively following links and instructions unless they are spelled out in specifics. Either way, let's get something in.

[BethGriggs]

Ah, probably superseded by #65493

[BethGriggs]

Now that the project’s AI use guidelines have landed (#62105), which permit the use of such tools, this proposes adding an AGENTS.md file to direct coding agents to the existing policies before they make changes.

Some tools may already discover and follow these documents, but this should make that behaviour slightly more deterministic. If it proves useful then we could consider extending with further instructions.

One consideration is that AGENTS.md is currently ignored (#62612). So, some contributors may have their own file which would subsequently conflict.

I know opinions vary, but the intent here is not to encourage usage of AI tools further than #62105, but to take the opportunity to guide the output.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.13%. Comparing base (21f0f27) to head (9e5a6ae).
⚠️ Report is 12 commits behind head on main.

Additional details and impacted files

@@ Coverage Diff @@## main #65463 +/- ##
==========================================
+ Coverage 90.12% 90.13% +0.01% 
==========================================
Files 752 751 -1 Lines 252315 252298 -17 Branches 47444 47437 -7 ==========================================
+ Hits 227395 227411 +16 + Misses 16217 16200 -17 + Partials 8703 8687 -16 

see 32 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[bmuenzenmeyer]

I'm tempted to adopt the convention of suggesting a CLAUDE.md that only states @AGENTS.md per many community workarounds - but I like more that we support the open standard only instead of supporting proprietary formats.

[MikeMcC399]

#65493 from @jasnell is taking a different approach, adding an AGENTS.md file with much more than just minimal contents. This would probably be preferable and it appears there is growing consensus on the content.

[jasnell]

I actually hadn't noticed this when I opened my alternative. I went with the more instructive version largely because I've noticed that some agents are... well, less inteligent than others when it comes to proactively following links and instructions unless they are spelled out in specifics. Either way, let's get something in.

[BethGriggs]

Ah, probably superseded by #65493

[BethGriggs]

Now that the project’s AI use guidelines have landed (#62105), which permit the use of such tools, this proposes adding an AGENTS.md file to direct coding agents to the existing policies before they make changes.

Some tools may already discover and follow these documents, but this should make that behaviour slightly more deterministic. If it proves useful then we could consider extending with further instructions.

One consideration is that AGENTS.md is currently ignored (#62612). So, some contributors may have their own file which would subsequently conflict.

I know opinions vary, but the intent here is not to encourage usage of AI tools further than #62105, but to take the opportunity to guide the output.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.13%. Comparing base (21f0f27) to head (9e5a6ae).
⚠️ Report is 12 commits behind head on main.

Additional details and impacted files

@@ Coverage Diff @@## main #65463 +/- ##
==========================================
+ Coverage 90.12% 90.13% +0.01% 
==========================================
Files 752 751 -1 Lines 252315 252298 -17 Branches 47444 47437 -7 ==========================================
+ Hits 227395 227411 +16 + Misses 16217 16200 -17 + Partials 8703 8687 -16 

see 32 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[bmuenzenmeyer]

I'm tempted to adopt the convention of suggesting a CLAUDE.md that only states @AGENTS.md per many community workarounds - but I like more that we support the open standard only instead of supporting proprietary formats.

[MikeMcC399]

#65493 from @jasnell is taking a different approach, adding an AGENTS.md file with much more than just minimal contents. This would probably be preferable and it appears there is growing consensus on the content.

[jasnell]

I actually hadn't noticed this when I opened my alternative. I went with the more instructive version largely because I've noticed that some agents are... well, less inteligent than others when it comes to proactively following links and instructions unless they are spelled out in specifics. Either way, let's get something in.

[BethGriggs]

Ah, probably superseded by #65493