chore: add AGENTS.md

[M0NsTeRRR]

Description

The purpose of this file is to provide specific, actionable instructions for AI agents and automated tooling to follow when contributing to this repository. This helps ensure that AI-assisted contributions align with the project's standards, conventions, and procedures.

This file was generated by the Gemini CLI under my supervision.

How Has This Been Tested?

Screenshots / Logs (if applicable)

Checklist:

• I have read and followed the contribution guidelines.
• Disclosed any use of AI (see our policy)
• I have updated the documentation accordingly.
• All new and existing tests passed.
• Successful build pnpm build
• Translation keys pnpm i18n:extract
• Database migration (if required)

Summary by CodeRabbit

• Documentation
  • Added comprehensive contributor guidelines documentation covering development setup procedures, build and test workflows, code style standards, pull request conventions including Conventional Commits, security vulnerability reporting, translation processes, and community code of conduct to support consistent project maintenance.

[coderabbitai]

📝 Walkthrough

Walkthrough

Adds a new AGENTS.md documentation file that provides guidelines for AI agents on contributing to the project, covering project overview, development setup, build and test procedures, database migrations, code style, translations, pull request conventions, security reporting, documentation structure, and code of conduct.

Changes

Cohort / File(s)	Summary
Documentation AGENTS.md	Introduces comprehensive contribution guidelines for AI agents, including development setup, testing procedures, code standards, pull request conventions with Conventional Commits guidance, and project policies.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~5 minutes

Possibly related PRs

• chore: update contributing guide regarding Automated AI Agent #2518: Modifies contribution documentation to add policies and guidance about automated AI agents and AI-assisted contributions.

Suggested reviewers

• gauthier-th
• fallenbagel

Poem

🐰 hops excitedly with clipboard
A guide for agents, bright and keen,
To contribute where code is seen,
With rules and paths both clear and neat,
Our AI friends make work complete! ✨

🚥 Pre-merge checks | ✅ 2

✅ Passed checks (2 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title 'chore: add AGENTS.md' directly and clearly summarizes the main change—adding a new documentation file for AI agents.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 2

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@AGENTS.md`:
- Line 72: Clarify that the example DB credentials in the AGENTS.md command
(DB_TYPE, DB_USER, DB_PASS) are only for local development and must not be used
in production; update the line containing "DB_TYPE=\"postgres\" DB_USER=postgres
DB_PASS=postgres pnpm migration:generate
server/migration/postgres/YourMigrationName" to include a short caveat or note
that these are local-dev defaults and recommend using secure,
environment-specific credentials (or a secrets manager) for non-local
environments.
- Around line 54-60: Update the CI workflow description to mention the project's
type-checking commands: add `pnpm typecheck` (and note the available variants
`typecheck:client` and `typecheck:server`) to the list of checks agents should
run locally and in CI (refer to the CI workflow summary and command list in
AGENTS.md); also rephrase the hedged sentence about tests to a direct statement
such as "This project does not include unit tests." so readers know there are no
unit test scripts to run.

---

ℹ️ Review info

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 55776ea and 549950a.

📒 Files selected for processing (1)

• AGENTS.md

[caillou]

I've personally never used an agent file. I have the feeling that it's very hard to not write things that are counter productive. Not that I have a lot of experience.

Recently there has been a study that seems to support this intuition:

https://news.ycombinator.com/item?id=47034087

Sometimes it helps, often it doesn't.

I leave this as a comment, this is just an intuition of mine. I especially have little experience with the agents...

Feel free to ignore.

[sudo-kraken]

I've personally never used an agent file. I have the feeling that it's very hard to not write things that are counter productive. Not that I have a lot of experience.

Recently there has been a study that seems to support this intuition:

https://news.ycombinator.com/item?id=47034087

Sometimes it helps, often it doesn't.

I leave this as a comment, this is just an intuition of mine. I especially have little experience with the agents...

Feel free to ignore.

I do agree with the sentiment, however people will continue to use agents to contribute regardless of what whether or not we publish the agents.md file. I think adding this in would at least enhance clarity on AI usage, style, and contribution standards so that it will at a minimum be in a stlye and standard aided by us.

[gauthier-th]

I've personally never used an agent file. I have the feeling that it's very hard to not write things that are counter productive. Not that I have a lot of experience.
Recently there has been a study that seems to support this intuition:
https://news.ycombinator.com/item?id=47034087
Sometimes it helps, often it doesn't.
I leave this as a comment, this is just an intuition of mine. I especially have little experience with the agents...
Feel free to ignore.

I do agree with the sentiment, however people will continue to use agents to contribute regardless of what whether or not we publish the agents.md file. I think adding this in would at least enhance clarity on AI usage, style, and contribution standards so that it will at a minimum be in a stlye and standard aided by us.

I think the main question is: do we want to allow AI agents for authoring the PR code? I'm fine with using it to explain the codebase, things like inline-suggestions, or ask some specific questions. But from what I reviewed here, most of the outputs produced by agents are low-quality, very verbose and touching a lot of stuff it shouldn't.

[sudo-kraken]

I've personally never used an agent file. I have the feeling that it's very hard to not write things that are counter productive. Not that I have a lot of experience.

Recently there has been a study that seems to support this intuition:

https://news.ycombinator.com/item?id=47034087

Sometimes it helps, often it doesn't.

I leave this as a comment, this is just an intuition of mine. I especially have little experience with the agents...

Feel free to ignore.

I do agree with the sentiment, however people will continue to use agents to contribute regardless of what whether or not we publish the agents.md file. I think adding this in would at least enhance clarity on AI usage, style, and contribution standards so that it will at a minimum be in a stlye and standard aided by us.

I've personally never used an agent file. I have the feeling that it's very hard to not write things that are counter productive. Not that I have a lot of experience.
Recently there has been a study that seems to support this intuition:
https://news.ycombinator.com/item?id=47034087
Sometimes it helps, often it doesn't.
I leave this as a comment, this is just an intuition of mine. I especially have little experience with the agents...
Feel free to ignore.

I do agree with the sentiment, however people will continue to use agents to contribute regardless of what whether or not we publish the agents.md file. I think adding this in would at least enhance clarity on AI usage, style, and contribution standards so that it will at a minimum be in a stlye and standard aided by us.

I think the main question is: do we want to allow AI agents for authoring the PR code? I'm fine with using it to explain the codebase, things like inline-suggestions, or ask some specific questions. But from what I reviewed here, most of the outputs produced by agents are low-quality, very verbose and touching a lot of stuff it shouldn't.

Agreed, then should we publish an update to our docs to state as much? that any PR from AI is prohibited but can be used to explain the codebase for those who wish to contribute?

[gauthier-th]

Agreed, then should we publish an update to our docs to state as much? that any PR from AI is prohibited but can be used to explain the codebase for those who wish to contribute?

I'd say something like: "PRs fully authored by AI agents are forbidden." The rest is fine.
WDYT @seerr-team/seerr-core?

[M0NsTeRRR]

For me it's already the case and covered by the first line of our contribution guide :

Automated AI-generated contributions without human review are not allowed and will be rejected

[Gylesie]

or even help with understanding the codebase.

I think this is too harsh or maybe said too broadly. Using AI to explain general concepts or some snippet of code should not need AI disclosure. Similar as to using a general search engine to do that or other analytical tools. And grasping the context and explaining things is what AIs excel at, anyway. And I don't think it is that helpful to know whether the person used AI to search the codebase or they did it manually - both of these actions are prone to misinterpretation or overlooking too.

The only exception that I could see is using an agentic AI crawling the entire codebase and drawing conclusions from there (compared to generative/assistive AI that requires continuous user input).

But this file is intended for agents anyway, so the wording might not matter that much here.

[gauthier-th]

For me it's already the case and covered by the first line of our contribution guide :

Automated AI-generated contributions without human review are not allowed and will be rejected

Fully authored AI-generated code with bad human review is still garbage. That's often what is happening.