(spark) branch master updated: [MINOR][DOC] Check both AGENTS.md and CLAUDE.md for nested project instructions

[yi...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

Yicong-Huang pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/spark.git


The following commit(s) were added to refs/heads/master by this push:
     new 158f853e994a [MINOR][DOC] Check both AGENTS.md and CLAUDE.md for nested project instructions
158f853e994a is described below

commit 158f853e994aa4bbdea59a3c2a71d7547b42b3ef
Author: Xiao Li <li...@databricks.com>
AuthorDate: Thu Aug 6 22:27:36 2026 +0000

    [MINOR][DOC] Check both AGENTS.md and CLAUDE.md for nested project instructions
    
    ### What changes were proposed in this pull request?
    
    Extends the "Project Instructions Discovery" section of the root `AGENTS.md` to cover `CLAUDE.md` as
    well as `AGENTS.md`.
    
    The section currently tells agents to look for nested `AGENTS.md` files only. That wording arrived in
    160b6193ee3 (`[MINOR][DOC] Use AGENTS.md for nested project instructions`), which replaced the
    previous `CLAUDE.md` wording. This PR:
    
    - says to check **both** names, and notes that a directory with only a `CLAUDE.md` still has
      instructions that must be read;
    - documents the convention this repository already follows at the root (`AGENTS.md` is the real file,
      `CLAUDE.md` is a symlink to it), so a reader knows either name resolves to the same content and
      there is nothing to reconcile;
    - states which name to use when adding instructions to a directory that has neither.
    
    Docs only -- no code, no build files.
    
    ### Why are the changes needed?
    
    `CLAUDE.md` is the filename Claude Code discovers by convention, and contributors add project
    instructions under that name. An agent that follows the current instruction literally looks for
    `AGENTS.md` only and skips a directory whose instructions live in a `CLAUDE.md`. Spelling out the
    symlink convention also keeps the two names from drifting into two separate documents.
    
    ### Does this PR introduce _any_ user-facing change?
    
    No.
    
    ### How was this patch tested?
    
    No tests; this changes prose in a markdown file.
    
    ### Was this patch authored or co-authored using generative AI tooling?
    
    Yes
    
    Closes #57825 from gatorsmile/agents-md-both-names.
    
    Authored-by: Xiao Li <li...@databricks.com>
    Signed-off-by: Yicong-Huang <17...@users.noreply.github.com>
---
 AGENTS.md | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/AGENTS.md b/AGENTS.md
index 36f1eccb02dd..fb6e064d348f 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -241,6 +241,8 @@ When exploring or working in any directory, always check for nested `AGENTS.md`
 directory and its ancestors. Read and follow every applicable file; instructions in a more specific
 directory take precedence for that directory's scope.
 
+A directory may carry these instructions as `CLAUDE.md` rather than `AGENTS.md`, so check **both** names — a directory with only a `CLAUDE.md` has project instructions you must read too. Where both files exist, one is typically the real file and the other is a symlink to it, although either name may be the real file. Reading either one is sufficient. When adding instructions to a directory that has neither file, create AGENTS.md as the real file and add a CLAUDE.md symlink beside it (ln  [...]
+
 ## Versioning and Branch Policy
 
 When a change needs a version — `@since` annotations, config `.version("...")` (`SQLConf` / `*Conf`), new `MimaExcludes` sections, etc. — use the version of the branch it first ships in, with `-SNAPSHOT` stripped. Determine that branch:


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@spark.apache.org
For additional commands, e-mail: commits-help@spark.apache.org