Entalpa vs CLAUDE.md and AGENTS.md
A file the agent can edit versus a store that refuses the edit. Everything else about this comparison is downstream of that.
What CLAUDE.md / AGENTS.md is
A plain markdown file at the root of the repository — CLAUDE.md for Claude Code, AGENTS.md for the thirty-odd agents that read it — describing the project, its conventions, and how to build and test it. It is loaded into the agent's context at the start of a session. It costs nothing, it is version-controlled with the code, and it is what most people should try first.
What CLAUDE.md / AGENTS.md does well
- Zero setup, zero cost, zero accounts, and no service that can go away.
- It lives in the repo, so it is reviewed, diffed, and rolled back with the code that it describes.
- Portable across agents by default — more than thirty read AGENTS.md, and the ones that do not can be pointed at it in one line.
- For a project small enough that one person holds it in their head, this is the whole answer and adding anything else is overhead.
Where the two differ
The agent can rewrite it
It is a file in a directory the agent has write access to. Agents edit it — sometimes helpfully, sometimes flattening a rule you spent an afternoon getting right. In Entalpa a locked requirement is filtered out of the set handed to the model before generation runs, so there is no rewrite to review.
Prose has no structure to trace
A paragraph saying 'refunds are approved by a manager above 200' cannot be linked to the story it came from, the requirement it constrains, or the acceptance criteria that test it. Entalpa stores stakeholders, stories and requirements as separate objects with the links between them, which is what makes 'what does this rule affect' a question with an answer.
Drift is silent
The file says what was true when someone last edited it. Nothing flags the gap between it and the code, and nobody notices until an agent confidently builds against the old rule.
Which one fits
Pick CLAUDE.md / AGENTS.md if
One repository, one or two people, a project whose rules fit on a page. The file wins and the answer stops there.
Pick Entalpa if
The file has started to drift, the agent has rewritten it, or you need to say what a rule affects — and the answer has to survive a session ending.
Questions this raises
Does Entalpa replace CLAUDE.md?
No, and it is not meant to. CLAUDE.md is a good place for build commands, style conventions and repository layout. Entalpa holds the domain rules — stakeholders, stories, requirements and the links between them — that the file cannot keep straight once there are enough of them.
Why does an agent edit CLAUDE.md at all?
Because it is a writable file in the working tree, and an agent asked to keep documentation current will treat it like any other. That is helpful right up until it rewrites a rule you meant to be fixed.
Try it against whatever you are using now
Connect Entalpa to the agent you already run, point it at a real project, and see whether a locked requirement set changes anything. Free to start, no card, and reading or writing your project over MCP never spends credits.
Compared with something else