AI guidelines#176
Conversation
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request adds a new section to the commit messages documentation regarding agent-generated commit messages, advising developers to write them manually or ensure they properly communicate intent. The review feedback highlights several grammatical, spelling, and formatting improvements, such as correcting 'adviced' to 'advised', hyphenating 'agent-generated', and improving sentence structure.
Important
The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.
| - [ ] `.agents/skills/agent-setup/SKILL.md`. Sets up MCP servers and skills for the used agent (ask agent to `invoke ".agents/skills/agent-setup/SKILL.md"`) | ||
|
|
||
| Some of these files will be pre-populated by our templates, make sure to adjust them for the project. | ||
|
|
There was a problem hiding this comment.
Maybe it's worth a note that we are experimenting on how to gather the best context for the project, and currently we are experimenting with Brewale and Palette as part of the process.
The art of gathering the right context will take a big part and will keep evolving a lot during the rest of 2026.
There was a problem hiding this comment.
| Gathering and sharing the right context is a big part of agentic programming, and this will keep evolving rapidly. We are experimenting with how to make useful tools, skills, and guidelines available for agents at the individual, project, and company level, while still leaving room for each project to define the business-specific context its agents need. At the time of writing, Brewale and Palette are part of that process. |
There was a problem hiding this comment.
I'm not really happy with the usage of the word "context" in my suggestion.
It is a complicated one because I'm mixing up two things:
- The context that is relevant in terms of the business
- "How we operate" which is closer to guidelines or commands
There was a problem hiding this comment.
Is it intentional that this paragraph starts with "- ". It will render a bullet list with a single bullet.
I'm not sure that I think this paragraph belongs here as part of project setup. It seems more like a "where we are - where we are heading" kinda thing. Or otherwise this would have to be a lot more actionable.
There was a problem hiding this comment.
Yeah, you are right. non-intentional. I was reading what was above.
Updated
|
|
||
| Use the following checklist to set up context for your project: | ||
|
|
||
| - [ ] `AGENTS.md` and symlink it from `CLAUDE.md` |
There was a problem hiding this comment.
| - [ ] `AGENTS.md` and symlink it from `CLAUDE.md` | |
| - [ ] `AGENTS.md` and symlink it from `CLAUDE.md` | |
| - [ ] `CLAUDE.md` Should operate as an index file indicating where to find information in the whole repo, including documentation, skills, or any other source of context that is external to the repo. |
There was a problem hiding this comment.
Not sure I agree on this, with this change it gets very fluffy whether or not AGENTS.md and CLAUDE.md are the same.
There was a problem hiding this comment.
The symlink line stays, so they are the same.
The line is addressing what Claude.md should contain. It is the single most important context file, so defining what goes there is important.
|
|
||
| On some projects we are experimenting with autonomous agents, letting them create and submit PRs on their own. | ||
|
|
||
| While this might save us some time, it also introduces security / economic concerns. Agents can easily get stuck in loops, and they are vulnerable to prompt injections. |
There was a problem hiding this comment.
The phrasing seems opinionated.
| While this might save us some time, it also introduces security / economic concerns. Agents can easily get stuck in loops, and they are vulnerable to prompt injections. | |
| As we experiment with new ways of working, we have to be aware of the security and economic concerns: agents can become stuck in loops, interpret instructions too literally, and are vulnerable to prompt injection. |
There was a problem hiding this comment.
Yeah, I was also not entirely happy with this sentence. The suggestion I however feel is becoming too wordy.
Could it instead be:
To avoid security and economic concerns, such as: agents getting stuck in loops, instructions interpretted too literally, and prompt injections, it is a requirement that autonomous agents:
[the list below]
Co-authored-by: Rune Philosof <57357936+runephilosof-abtion@users.noreply.github.com>
Co-authored-by: Hari Abtion <hc@abtion.com>
Co-authored-by: Hari Abtion <hc@abtion.com>
Co-authored-by: Hari Abtion <hc@abtion.com>
3 participants
No description provided.