AGENTS.md: Add file
PATCH 1/1 AGENTS.md: Add file
Adds an AGENTS.md file to direct agents about better commit practices. This is specifically *not* attempting to improve or direct agents about the code that they generate. It has been noted that when AI is used to write code and commit code, the commit messages that it generates can be long and draw out (e.g. a "wall of text") which affects maintainer ability to review the patches. Therefore, this AGENTS.md text file is only focused on instructing the agents to produce commits that are more suitable for our review process.
This file is intended to help agents comply with the existing project document (and specifically, the AI Policy[1]), and does not constitute a new policy. Any subsequent changes to this file should derive and comply with the project documentation and policies.
It what can only be described as the most bizarre interaction with a computer I've ever experienced, the best way to get an AGENTS.md file that AI agents actually respect as intended is to instruct one to write it for you (as opposed to writing one by hand). Future edits to this file will likely be most effective if written by an Agent.
[1]: https://docs.yoctoproject.org/contributor-guide/submit-changes.html#acceptance-of-ai-generated-code
AI-Generated: Uses Claude Sonnet 5
--- AGENTS.md | 55 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 55 insertions(+) create mode 100644 AGENTS.md
AGENTS.md new file +55 -0
@@ -0,0 +1,55 @@ +# AGENTS.md — OpenEmbedded-Core & BitBake + +Guidance for AI agents working in the `openembedded-core` repository. + +## Making changes + +- No web PR workflow. The project use **patches on mailing lists** + ([email protected]) tracked via patchwork. Don't + assume GitHub PR conventions apply. +- One Git commit per logical change (e.g., one commit per recipe added/upgraded). +- Commit message format: + - Summary line prefixed with the recipe name or short file path, + e.g. `openssl: upgrade to 3.2.1` or `classes/rootfs-postcommands: fix typo`. + Check `git log --oneline <path>` for the prefix convention already in use. + - Body explains what/why/how, and how it was tested. + - Prefer explaining *why* a change was made over restating what's + already visible in the diff, unless the diff is long/hard to follow + or the reason for the change isn't known — in those cases, summarize + what changed instead. + - Be succinct: use as few sentences and lines as possible while still + covering what/why/how. Cut restatements, filler, and any detail a + reviewer wouldn't need (e.g. exact test-run counts, step-by-step + narration). One short paragraph is usually enough; only add a second + if a distinct point (e.g. testing) needs it. + - The most relevant information about the change should be presented first. + - Reference bugs as `Fixes [YOCTO #1234]` when applicable. + - Use `Reported-by`, `Suggested-by`, `Tested-by`, `Reviewed-by`, `Cc` tags + where relevant. + +### AI-generated code + +- **Mandatory**: any AI-generated code/commit must add an `AI-Generated:` + tag line in the commit message, placed *before* `Signed-off-by` (if present), + naming the tool used, e.g.: + + ``` + component: Add the ability to ... + + AI-Generated: Uses GitHub Copilot + ``` + +- Also add a code comment or other in-patch indication that the code is + AI-generated (recommended, not just the commit tag). +- **Mandatory** Do not automatically add a `Signed-off-by`. This tag must + always be written by the human contributor, never by the AI tool — the + DCO/Developer's Statement of Origin still applies in full. +- Contributor is responsible for confirming the AI tool's terms don't + conflict with the project's open source license/IP policy, and that + any third-party copyrighted material in the AI output is properly + licensed before contributing it. See the Linux Foundation generative AI + guidance: https://www.linuxfoundation.org/legal/generative-ai. +- `Co-authored-by` should be omitted in favor of the `AI-generated` tag. + +When in doubt about a convention, grep `git log --oneline <path>` in the +target repo for precedent before inventing a new style.
Signature · 1 line
2.54.0
Quoted 60 lines
-----Original Message-----
From: [email protected] <openembedded- [email protected]> On Behalf Of Joshua Watt via lists.openembedded.org
Sent: Monday, August 10, 2026 6:12 PM
To: [email protected]
Cc: Joshua Watt <[email protected]>
Subject: [OE-core][PATCH] AGENTS.md: Add fileAdds an AGENTS.md file to direct agents about better commit practices. This is specifically *not* attempting to improve or direct agents about the code that they generate. It has been noted that when AI is used to write code and commit code, the commit messages that it generates can be long and draw out (e.g. a "wall of text") which affects maintainer ability to review the patches. Therefore, this AGENTS.md text file is only focused on instructing the agents to produce commits that are more suitable for our review process.
This file is intended to help agents comply with the existing project document (and specifically, the AI Policy[1]), and does not constitute a new policy. Any subsequent changes to this file should derive and comply with the project documentation and policies.
It what can only be described as the most bizarre interaction with a computer I've ever experienced, the best way to get an AGENTS.md file that AI agents actually respect as intended is to instruct one to write it for you (as opposed to writing one by hand). Future edits to this file will likely be most effective if written by an Agent.
[1]: https://docs.yoctoproject.org/contributor-guide/submit-
changes.html#acceptance-of-ai-generated-code
AI-Generated: Uses Claude Sonnet 5Signed-off-by Joshua Watt <[email protected]>--- AGENTS.md | 55 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 55 insertions(+) create mode 100644 AGENTS.mdAGENTS.md new file +10
-0@@ -0,0 +1,55 @@ +# AGENTS.md — OpenEmbedded-Core & BitBake + +Guidance for AI agents working in the `openembedded-core` repository. + +## Making changes + +- No web PR workflow. The project use **patches on mailing lists** + ([email protected]) tracked via patchwork. Don't + assume GitHub PR conventions apply. +- One Git commit per logical change (e.g., one commit per recipeadded/upgraded).
+- Commit message format:
AGENTS.md line 13
Quoted 2 lines
AGENTS.md new file +2
-0+ - Summary line prefixed with the recipe name or short file path, + e.g. `openssl: upgrade to 3.2.1` or `classes/rootfs-postcommands: fix typo`.
Current standard for upgrade commit message is "openssl: upgrade 3.2.0 -> 3.2.1"
Quoted 5 lines
AGENTS.md new file +5
-0+ Check `git log --oneline <path>` for the prefix convention already in use. + - Body explains what/why/how, and how it was tested. + - Prefer explaining *why* a change was made over restating what's + already visible in the diff, unless the diff is long/hard to follow + or the reason for the change isn't known — in those cases, summarize
AGENTS.md line 28
Quoted 10 lines
AGENTS.md new file +10
-0+ what changed instead. + - Be succinct: use as few sentences and lines as possible while still + covering what/why/how. Cut restatements, filler, and any detail a + reviewer wouldn't need (e.g. exact test-run counts, step-by-step + narration). One short paragraph is usually enough; only add a second + if a distinct point (e.g. testing) needs it. + - The most relevant information about the change should be presented first. + - Reference bugs as `Fixes [YOCTO #1234]` when applicable. + - Use `Reported-by`, `Suggested-by`, `Tested-by`, `Reviewed-by`, `Cc` tags + where relevant.
Where should the agent get yocto bugzilla ID or the personal tags from? This may lead to hallucinations.
Maybe we can suggest to ask/remind user about those?
Quoted 14 lines
AGENTS.md new file +14
-0+ +### AI-generated code + +- **Mandatory**: any AI-generated code/commit must add an `AI-Generated:` + tag line in the commit message, placed *before* `Signed-off-by` (if present), + naming the tool used, e.g.: + + ``` + component: Add the ability to ... + + AI-Generated: Uses GitHub Copilot + ``` + +- Also add a code comment or other in-patch indication that the code is
AGENTS.md line 52
Quoted 10 lines
AGENTS.md new file +10
-0+ AI-generated (recommended, not just the commit tag). +- **Mandatory** Do not automatically add a `Signed-off-by`. This tag must + always be written by the human contributor, never by the AI tool — the + DCO/Developer's Statement of Origin still applies in full. +- Contributor is responsible for confirming the AI tool's terms don't + conflict with the project's open source license/IP policy, and that + any third-party copyrighted material in the AI output is properly + licensed before contributing it. See the Linux Foundation generative AI + guidance: https://www.linuxfoundation.org/legal/generative-ai. +- `Co-authored-by` should be omitted in favor of the `AI-generated` tag.
Do you have references about AI-generated git tags?
I think that the AssistedBy is the current state-of-the-art way of annotating this...
Quoted 5 lines
AGENTS.md new file +3
-1+ +When in doubt about a convention, grep `git log --oneline <path>` in the +target repo for precedent before inventing a new style. --2.54.0
Quoted 104 lines · Peter Marko via lists.openembedded.org
Quoted 60 lines
-----Original Message-----
From: [email protected] <openembedded- [email protected]> On Behalf Of Joshua Watt via lists.openembedded.org
Sent: Monday, August 10, 2026 6:12 PM
To: [email protected]
Cc: Joshua Watt <[email protected]>
Subject: [OE-core][PATCH] AGENTS.md: Add fileAdds an AGENTS.md file to direct agents about better commit practices. This is specifically *not* attempting to improve or direct agents about the code that they generate. It has been noted that when AI is used to write code and commit code, the commit messages that it generates can be long and draw out (e.g. a "wall of text") which affects maintainer ability to review the patches. Therefore, this AGENTS.md text file is only focused on instructing the agents to produce commits that are more suitable for our review process.
This file is intended to help agents comply with the existing project document (and specifically, the AI Policy[1]), and does not constitute a new policy. Any subsequent changes to this file should derive and comply with the project documentation and policies.
It what can only be described as the most bizarre interaction with a computer I've ever experienced, the best way to get an AGENTS.md file that AI agents actually respect as intended is to instruct one to write it for you (as opposed to writing one by hand). Future edits to this file will likely be most effective if written by an Agent.
[1]: https://docs.yoctoproject.org/contributor-guide/submit-
changes.html#acceptance-of-ai-generated-code
AI-Generated: Uses Claude Sonnet 5Signed-off-by Joshua Watt <[email protected]>--- AGENTS.md | 55 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 55 insertions(+)create mode 100644 AGENTS.md
AGENTS.md new file +10
-0@@ -0,0 +1,55 @@ +# AGENTS.md — OpenEmbedded-Core & BitBake + +Guidance for AI agents working in the `openembedded-core` repository. + +## Making changes + +- No web PR workflow. The project use **patches on mailing lists** + ([email protected]) tracked via patchwork. Don't + assume GitHub PR conventions apply. +- One Git commit per logical change (e.g., one commit per recipeadded/upgraded).
+- Commit message format:AGENTS.md line 13
Quoted 2 lines
AGENTS.md new file +2
-0+ - Summary line prefixed with the recipe name or short file path, + e.g. `openssl: upgrade to 3.2.1` or `classes/rootfs-postcommands: fix typo`.Current standard for upgrade commit message is "openssl: upgrade 3.2.0 -> 3.2.1"
Quoted 5 lines
AGENTS.md new file +5
-0+ Check `git log --oneline <path>` for the prefix convention already in use. + - Body explains what/why/how, and how it was tested. + - Prefer explaining *why* a change was made over restating what's + already visible in the diff, unless the diff is long/hard to follow + or the reason for the change isn't known — in those cases, summarizeAGENTS.md line 28
Quoted 10 lines
AGENTS.md new file +10
-0+ what changed instead. + - Be succinct: use as few sentences and lines as possible while still + covering what/why/how. Cut restatements, filler, and any detail a + reviewer wouldn't need (e.g. exact test-run counts, step-by-step + narration). One short paragraph is usually enough; only add a second + if a distinct point (e.g. testing) needs it. + - The most relevant information about the change should be presented first. + - Reference bugs as `Fixes [YOCTO #1234]` when applicable. + - Use `Reported-by`, `Suggested-by`, `Tested-by`, `Reviewed-by`, `Cc` tags + where relevant.Where should the agent get yocto bugzilla ID or the personal tags from? This may lead to hallucinations.
Maybe we can suggest to ask/remind user about those?Quoted 19 lines
AGENTS.md new file +19
-0+ +### AI-generated code + +- **Mandatory**: any AI-generated code/commit must add an `AI-Generated:` + tag line in the commit message, placed *before* `Signed-off-by` (if present), + naming the tool used, e.g.: + + ``` + component: Add the ability to ... + + AI-Generated: Uses GitHub Copilot + ``` + +- Also add a code comment or other in-patch indication that the code is + AI-generated (recommended, not just the commit tag). +- **Mandatory** Do not automatically add a `Signed-off-by`. This tag must + always be written by the human contributor, never by the AI tool — the + DCO/Developer's Statement of Origin still applies in full. +- Contributor is responsible for confirming the AI tool's terms don't
Quoted 9 lines
AGENTS.md line 52
Quoted 5 lines
AGENTS.md new file +5
-0+ conflict with the project's open source license/IP policy, and that + any third-party copyrighted material in the AI output is properly + licensed before contributing it. See the Linux Foundation generative AI + guidance: https://www.linuxfoundation.org/legal/generative-ai. +- `Co-authored-by` should be omitted in favor of the `AI-generated` tag.Do you have references about AI-generated git tags?
I think that the AssistedBy is the current state-of-the-art way of annotating this...
https://docs.yoctoproject.org/contributor-guide/submit-changes.html#acceptance-of-ai-generated-code
Quoted 63 lines · Peter
Quoted 48 lines
-----Original Message-----
From: [email protected] <openembedded- [email protected]> On Behalf Of Joshua Watt via lists.openembedded.org
Sent: Monday, August 10, 2026 6:12 PM
To: [email protected]
Cc: Joshua Watt <[email protected]>
Subject: [OE-core][PATCH] AGENTS.md: Add fileAdds an AGENTS.md file to direct agents about better commit practices. This is specifically *not* attempting to improve or direct agents about the code that they generate. It has been noted that when AI is used to write code and commit code, the commit messages that it generates can be long and draw out (e.g. a "wall of text") which affects maintainer ability to review the patches. Therefore, this AGENTS.md text file is only focused on instructing the agents to produce commits that are more suitable for our review process.
This file is intended to help agents comply with the existing project document (and specifically, the AI Policy[1]), and does not constitute a new policy. Any subsequent changes to this file should derive and comply with the project documentation and policies.
It what can only be described as the most bizarre interaction with a computer I've ever experienced, the best way to get an AGENTS.md file that AI agents actually respect as intended is to instruct one to write it for you (as opposed to writing one by hand). Future edits to this file will likely be most effective if written by an Agent.
[1]: https://docs.yoctoproject.org/contributor-guide/submit-
changes.html#acceptance-of-ai-generated-code
AI-Generated: Uses Claude Sonnet 5Signed-off-by Joshua Watt <[email protected]>--- AGENTS.md | 55 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 55 insertions(+) create mode 100644 AGENTS.mdAGENTS.md new file +0
-0AGENTS.md line 8
Quoted 8 lines
AGENTS.md new file +8
-0@@ -0,0 +1,55 @@ +# AGENTS.md — OpenEmbedded-Core & BitBake + +Guidance for AI agents working in the `openembedded-core` repository. + +## Making changes + +- No web PR workflow. The project use **patches on mailing lists** + ([email protected]) tracked via patchwork.Don't
AGENTS.md
Quoted 6 lines
AGENTS.md new file +2
-0+ assume GitHub PR conventions apply. +- One Git commit per logical change (e.g., one commit per recipeadded/upgraded).
+- Commit message format:AGENTS.md new file +2
-0+ - Summary line prefixed with the recipe name or short file path, + e.g. `openssl: upgrade to 3.2.1` or `classes/rootfs-postcommands:
Quoted 4 lines
fix typo`.
Current standard for upgrade commit message is "openssl: upgrade 3.2.0 -> 3.2.1"
I would probably recommend leaving it and seeing how it goes? This file is almost certainly better written reactively to things we actually see happening than trying to guess what an agent _might_ do. It's really hard to tell how the Agent will behave with regard to some of these things, and I tried pretty hard to not try and enumerate all the ways it _could_ go wrong (e.g. Do we need to specify that commit messages should not contain emoji? I just don't know). Instead, we plan to keep an eye on it and if we see agents doing other things that are undesirable, then we will add more rules to this file as appropriate.
FWIW, most of this section was generated by Claude by reading our documentation, although I did specifically prompt it to add in text about making more succinct commit messages
Quoted 15 lines
Quoted 2 lines
+ Check
git log --oneline <path>for the prefix convention alreadyin use.
AGENTS.md
Quoted 11 lines
AGENTS.md new file +11
-0+ - Body explains what/why/how, and how it was tested. + - Prefer explaining *why* a change was made over restating what's + already visible in the diff, unless the diff is long/hard to follow + or the reason for the change isn't known — in those cases, summarize + what changed instead. + - Be succinct: use as few sentences and lines as possible while still + covering what/why/how. Cut restatements, filler, and any detail a + reviewer wouldn't need (e.g. exact test-run counts, step-by-step + narration). One short paragraph is usually enough; only add a second + if a distinct point (e.g. testing) needs it. + - The most relevant information about the change should be presentedfirst.
Quoted 8 lines
AGENTS.md
Quoted 2 lines
AGENTS.md new file +2
-0+ - Reference bugs as `Fixes [YOCTO #1234]` when applicable. + - Use `Reported-by`, `Suggested-by`, `Tested-by`, `Reviewed-by`, `Cc`tags
Quoted 1 line
+ where relevant.
Where should the agent get yocto bugzilla ID or the personal tags from? This may lead to hallucinations.
Maybe we can suggest to ask/remind user about those?
If we really think an agent will trip on this, we should just remove the line for now until we have actual examples of it doing something wrong (which, I understand, is a little tricky, because we also expect the human submitting the patch to look at and fix any problems :)
In the future if we did change anything, I'd probably rather err on the side of asking the user or never adding a tag than trying to look up a bug in bugzilla. The agents are pretty good at incorporating context, so a prompt like, "Fix Yocto bug 12345 by doing XYZ" would likely do the right thing.
Quoted 40 lines
AGENTS.md
Quoted 4 lines
AGENTS.md new file +4
-0+ +### AI-generated code + +- **Mandatory**: any AI-generated code/commit must add an
AI-Generated:Quoted 1 line
+ tag line in the commit message, placed *before*
Signed-off-by(ifpresent),
AGENTS.md
Quoted 11 lines
AGENTS.md new file +11
-0+ naming the tool used, e.g.: + + ``` + component: Add the ability to ... + + AI-Generated: Uses GitHub Copilot + ``` + +- Also add a code comment or other in-patch indication that the code is + AI-generated (recommended, not just the commit tag). +- **Mandatory** Do not automatically add a `Signed-off-by`. This tagmust
AGENTS.md
Quoted 6 lines
AGENTS.md new file +6
-0+ always be written by the human contributor, never by the AI tool — the + DCO/Developer's Statement of Origin still applies in full. +- Contributor is responsible for confirming the AI tool's terms don't + conflict with the project's open source license/IP policy, and that + any third-party copyrighted material in the AI output is properly + licensed before contributing it. See the Linux Foundation generativeAI
Quoted 2 lines
+ guidance: https://www.linuxfoundation.org/legal/generative-ai.
+-Co-authored-byshould be omitted in favor of theAI-generatedtag.Do you have references about AI-generated git tags?
I think that the AssistedBy is the current state-of-the-art way of annotating this...Quoted 5 lines
AGENTS.md new file +3
-1+ +When in doubt about a convention, grep `git log --oneline <path>` in the +target repo for precedent before inventing a new style. --2.54.0
PATCH v2 1/1 AGENTS.md: Add file
In reply to Joshua WattAdds an AGENTS.md file to direct agents about better commit practices. This is specifically *not* attempting to improve or direct agents about the code that they generate. It has been noted that when AI is used to write code and commit code, the commit messages that it generates can be long and draw out (e.g. a "wall of text") which affects maintainer ability to review the patches. Therefore, this AGENTS.md text file is only focused on instructing the agents to produce commits that are more suitable for our review process.
This file is intended to help agents comply with the existing project document (and specifically, the AI Policy[1]), and does not constitute a new policy. Any subsequent changes to this file should derive and comply with the project documentation and policies.
It what can only be described as the most bizarre interaction with a computer I've ever experienced, the best way to get an AGENTS.md file that AI agents actually respect as intended is to instruct one to write it for you (as opposed to writing one by hand). Future edits to this file will likely be most effective if written by an Agent.
[1]: https://docs.yoctoproject.org/contributor-guide/submit-changes.html#acceptance-of-ai-generated-code
AI-Generated: Uses Claude Sonnet 5
--- AGENTS.md | 55 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 55 insertions(+) create mode 100644 AGENTS.md
AGENTS.md new file +55 -0
@@ -0,0 +1,55 @@ +# AGENTS.md — OpenEmbedded-Core & BitBake + +Guidance for AI agents working in the `openembedded-core` repository. + +## Making changes + +- No web PR workflow. The project use **patches on mailing lists** + ([email protected]) tracked via patchwork. Don't + assume GitHub PR conventions apply. +- One Git commit per logical change (e.g., one commit per recipe added/upgraded). +- Commit message format: + - Summary line prefixed with the recipe name or short file path, + e.g. `openssl: upgrade 3.2.0 -> 3.2.1` or `classes/rootfs-postcommands: fix typo`. + Check `git log --oneline <path>` for the prefix convention already in use. + - Body explains what/why/how, and how it was tested. + - Prefer explaining *why* a change was made over restating what's + already visible in the diff, unless the diff is long/hard to follow + or the reason for the change isn't known — in those cases, summarize + what changed instead. + - Be succinct: use as few sentences and lines as possible while still + covering what/why/how. Cut restatements, filler, and any detail a + reviewer wouldn't need (e.g. exact test-run counts, step-by-step + narration). One short paragraph is usually enough; only add a second + if a distinct point (e.g. testing) needs it. + - The most relevant information about the change should be presented first. + - Reference bugs as `[YOCTO #1234]` when applicable. + - Use `Reported-by`, `Suggested-by`, `Tested-by`, `Reviewed-by`, `Cc` tags + where relevant. + +### AI-generated code + +- **Mandatory**: any AI-generated code/commit must add an `AI-Generated:` + tag line in the commit message, placed *before* `Signed-off-by` (if present), + naming the tool used, e.g.: + + ``` + component: Add the ability to ... + + AI-Generated: Uses GitHub Copilot + ``` + +- Also add a code comment or other in-patch indication that the code is + AI-generated (recommended, not just the commit tag). +- **Mandatory** Do not automatically add a `Signed-off-by`. This tag must + always be written by the human contributor, never by the AI tool — the + DCO/Developer's Statement of Origin still applies in full. +- Contributor is responsible for confirming the AI tool's terms don't + conflict with the project's open source license/IP policy, and that + any third-party copyrighted material in the AI output is properly + licensed before contributing it. See the Linux Foundation generative AI + guidance: https://www.linuxfoundation.org/legal/generative-ai. +- `Co-authored-by` should be omitted in favor of the `AI-generated` tag. + +When in doubt about a convention, grep `git log --oneline <path>` in the +target repo for precedent before inventing a new style.
Signature · 1 line
2.55.0