Free tools Windows power users keep installed
One-click scans. No signup required.
For a repository used by Codex, Claude Code, and compatible coding agents, use AGENTS.md as the shared project guidance file where those tools support it. Add CLAUDE.md only if Claude needs separate instructions or its deployed configuration does not load AGENTS.md. If you use both, make Claude’s behavior explicit: configure it to read both, or import the shared file from CLAUDE.md.
What is the difference between AGENTS.md and CLAUDE.md?
They are repository instruction files, but their names are not interchangeable in every coding-agent setup. AGENTS.md is the practical shared default for Codex and compatible tools. CLAUDE.md is Claude Code’s tool-specific instruction file. Which files an agent actually reads depends on its harness, version, settings, and file-discovery rules.
As an Amazon Associate I earn from qualifying purchases.
OpenAI’s Codex documentation says, “Codex reads AGENTS.md files before doing any work.” Codex can also use configured fallback filenames, but a differently named file is not automatically recognized just because another tool supports it. OpenAI’s AGENTS.md documentation
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Which file should you use for a shared repository?
Use one AGENTS.md as the source of truth for instructions that apply across agents: project conventions, build and test commands, code organization, and task-specific expectations. This reduces the risk that two files drift apart and give agents conflicting guidance.
Claude Code can read AGENTS.md directly in current documented configurations, so a separate CLAUDE.md is not automatically required. But the feature is version- and configuration-dependent: Anthropic documents direct support for Claude Code v2.1.277 or later, with exceptions such as older versions or a disabled built-in plugin. Check the version and settings used by your team before relying on it. Anthropic’s Claude Code memory documentation
What happens when both files exist?
Claude Code’s documented default mode, claude-md-or-agents-md, reads CLAUDE.md files, or falls back to AGENTS.md when there is no CLAUDE.md or CLAUDE.local.md in the working directory or above it. In this mode, adding a CLAUDE.md can therefore affect whether Claude loads the shared file.
Rank #2
- New design has wider shelves and supports, increasing stability for wide books. Shelf width is now 14.5".
- Easily holds two large medical coding books.
- Made in the USA - Minor assembly required.
Anthropic documents two other modes: claude-md-and-agents-md reads both, with Claude files before AGENTS files at each directory; claude-md reads only Claude files. A managed-only mode limits launch-time project and user instruction loading to managed Claude instructions and auto memory. Confirm the active configuration rather than assuming that both files are being combined. Anthropic’s Claude Code memory documentation
Keep one source of truth and add Claude-only guidance
If Claude needs extra instructions, Anthropic documents putting @AGENTS.md at the top of CLAUDE.md, then adding Claude-specific guidance below it. A symlink is another documented option. This approach keeps shared rules in one place while allowing the Claude file to carry tool-specific content.
Use two independent files only deliberately
If both files contain overlapping rules, assign a clear purpose to each and tell the team which one governs shared conventions. Otherwise, a change made to one file may leave the other outdated. Also account for Claude’s documented behavior for hooks and some additional directories if your setup uses those features; the main project-file rules do not necessarily describe every instruction-loading mechanism.
How do discovery and nested instructions differ?
File format alone does not guarantee that every harness discovers instructions the same way. Codex starts with global and project instructions. For project scope, it starts at the project root—typically the Git root—and walks down to the current working directory. In each directory, it checks AGENTS.override.md, then AGENTS.md, then configured fallback filenames, loading at most one file per directory. A nearer override can specialize guidance; discovery stops at the current directory. OpenAI’s AGENTS.md documentation
Rank #4
- Students build unmatched deductive-reasoning skills as they become crime-solving stars
- Most scenarios have more than one plausible outcome, allowing individuals or groups to broadly interpret evidence
- Includes interpretive handwriting, body language, fingerprinting, and many more activities
Claude Code’s file choice and ordering depend on its configured memory mode, while VS Code support depends on the selected harness and settings. Microsoft lists AGENTS.md as a recommended project instruction format for OpenAI Codex and Copilot, and CLAUDE.md for the Anthropic Claude harness. It recommends a shared supported format where possible, while noting that nested discovery and support depend on the harness and configuration. Microsoft’s VS Code custom-instructions documentation
| Tool or setup | What the documentation establishes | What to verify |
|---|---|---|
| Codex | Reads AGENTS.md; supports a project hierarchy and configured fallback filenames. |
Whether any nonstandard filename is configured as a fallback. |
| Claude Code | Can read AGENTS files in supported current configurations; its mode determines whether it reads Claude files, AGENTS files, or both. | Version, settings, plugin availability, and the active memory mode. |
| VS Code with an agent harness | Documents AGENTS.md for Codex and Copilot, and CLAUDE.md for Claude. | Selected harness, settings, and nested-file discovery. |
How should you write the instructions?
Make the file useful at the moment an agent begins work. Keep durable, repository-specific guidance there; move long procedures or specialized task guidance to the appropriate tool feature when available. Avoid treating the instruction file as a dumping ground for every possible rule.
Best Value
- Used Book in Good Condition
- State essential commands and prerequisites clearly, including how to run the relevant tests.
- Use concise, actionable rules tied to this repository rather than broad advice that applies to any project.
- Separate truly tool-specific requirements from shared project rules, and explain the relationship between files.
- Check nested instructions and overrides in each harness so a local rule does not silently diverge from the root guidance.
- Review the instructions when project tooling or conventions change, just as you would maintain build documentation.
What do studies say about context files?
Two 2026 studies offer useful cautions, not universal verdicts. Prakhar Khatri’s two-agent ablation study evaluated 288 runs across 17 tasks using Claude Code and Codex. It reported no measurable correctness change from context strategy within the study’s equivalence bounds. That result is limited to the agents, tasks, and conditions studied; it does not establish that instruction files never help. Study: Do Context Files Help Coding Agents?
A separate study by Helio Victor F. dos Santos and coauthors analyzed 100 popular open-source repositories containing AGENTS.md or CLAUDE.md. Using the authors’ heuristics, it identified at least one configuration smell in 91 repositories: lint leakage in 62% of sampled files, context bloat in 42%, and skill leakage in 35%. These are findings from that sample and method—not prevalence estimates for all repositories. The practical lesson is to keep instruction files scoped and maintainable, not to assume a particular file name guarantees better results. Study: Configuration Smells in AGENTS.md Files
Quick Recap
A practical decision rule
- Start with
AGENTS.md. Put shared project guidance there for Codex and other tools that support the format. - Check Claude Code’s actual setup. Confirm its version, settings, and memory mode before assuming it reads AGENTS files.
- Add
CLAUDE.mdonly for a reason. Use it for Claude-specific rules or when the deployed configuration requires it. - Make the relationship explicit. Configure Claude to read both, or use the documented
@AGENTS.mdimport or a symlink so the shared instructions stay synchronized. - Validate nested behavior per harness. Do not assume overrides, fallback names, or directory discovery work identically across tools.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
