refactor(agents-init): extract discovery prune set into _agents_init_find

functions/_agents_init_find.fish

@@ -0,0 +1,44 @@
+# Copyright (C) 2026 Rootiest
+# SPDX-License-Identifier: AGPL-3.0-or-later
+
+# SYNOPSIS
+#   _agents_init_find <root> <find-expression>...
+#
+# DESCRIPTION
+#   Runs find over <root> with the prune set agents-init discovery uses,
+#   then applies <find-expression> to every entry that survives. Pruned:
+#   dot-directories (.git, .claude, .github -- tool state, not scoped
+#   project dirs), any AGENTS/ (a mirror, never a source), node_modules,
+#   generated-output directories (build, dist, out, target -- an
+#   instruction file there is a build artifact), and nested repositories,
+#   submodules and worktrees (their own .git marks another project).
+#   -mindepth 1 keeps <root> itself, which has a .git, from pruning the
+#   whole walk.
+#
+#   Shared by agents-init (instruction-file discovery) and agents-cleanup
+#   (symlinks into AGENTS/), so the two never disagree about which part of
+#   the tree belongs to the project.
+#
+# ARGUMENTS
+#   root             Absolute path to the project root
+#   find-expression  find primaries applied to every unpruned entry; must
+#                    carry its own action (for example -print)
+#
+# EXIT STATUS
+#   0  find completed
+#   1  No root given, or find failed
+#
+# RETURNS
+#   Whatever <find-expression> prints.
+#
+# EXAMPLE
+#   _agents_init_find /path/to/project \( -name AGENTS.md -o -name CLAUDE.md \) -print
+#   _agents_init_find /path/to/project -type l -print
+function _agents_init_find --argument-names root
+    test -n "$root"; or return 1
+    find "$root" -mindepth 1 \
+        -type d \( -name '.*' -o -name AGENTS -o -name node_modules \
+        -o -name build -o -name dist -o -name out -o -name target \
+        -o -exec test -e '{}/.git' \; \) -prune -o \
+        $argv[2..]
+end

functions/agents-init.fish

@@ -5,7 +5,7 @@
 #   12-ai-and-developer-tools
 #
 # DEPENDENCIES
-#   _agents_init_sync_instructions, _agents_repo_install_tools, _agents_repo_sync, _agents_init_ensure_gitignore
+#   _agents_init_find, _agents_init_sync_instructions, _agents_repo_install_tools, _agents_repo_sync, _agents_init_ensure_gitignore
 #
 # CLASSIFICATION
 #   self-limiting(rm,mkdir,grep), bypasses-shadow(mv), manual-section(16-agent-tooling)
@@ -238,22 +238,11 @@ function agents-init --description 'scaffold AGENTS/ sub-repo with agent spec fi
         #
         # Discovery stays inside this project: a non-git root (a lone
         # agent file in, say, ~) syncs only itself -- walking it would
-        # reach into every unrelated tree below. In a git root, pruned:
+        # reach into every unrelated tree below. In a git root the walk
-        # any AGENTS/ (a mirror, never a source), dot-directories (.git,
+        # uses the shared prune set; see _agents_init_find.
-        # .claude, .github: tool state, not scoped project dirs),
-        # node_modules, generated-output directories (build, dist, out,
-        # target: an instruction file there is a build artifact, never a
-        # source -- pruned outright, before tracked-file protection would
-        # even be consulted), and nested repos/submodules/worktrees (their
-        # own .git marks another project). -mindepth 1 keeps the root
-        # itself, which has a .git, from pruning the whole walk.
         set -l found
         if test $in_git -eq 1
-            set found (find "$root" -mindepth 1 \
+            set found (_agents_init_find "$root" \( -name AGENTS.md -o -name CLAUDE.md \) -print)
-                -type d \( -name '.*' -o -name AGENTS -o -name node_modules \
-                -o -name build -o -name dist -o -name out -o -name target \
-                -o -exec test -e '{}/.git' \; \) -prune -o \
-                \( -name AGENTS.md -o -name CLAUDE.md \) -print)
         end
         set -l rels "."
         for f in $found

tests/test-agents-init.fish

@@ -461,5 +461,20 @@ _agents_init_path_is_protected $g1 "$g1/a[1]/AGENTS.md"
 set -l protected $status
 check "glob false-match: untracked a[1]/AGENTS.md is not protected" 1 "$protected"
 
+echo ""
+echo "== _agents_init_find: shared prune set =="
+
+set -l pf (new_repo)
+mkdir -p $pf/src $pf/node_modules/pkg $pf/.claude $pf/AGENTS $pf/build $pf/nested
+git -C $pf/nested init -q
+ln -s ../x $pf/src/link
+ln -s x $pf/node_modules/pkg/link
+ln -s x $pf/.claude/link
+ln -s x $pf/AGENTS/link
+ln -s x $pf/build/link
+ln -s x $pf/nested/link
+check "find: only the unpruned symlink" $pf/src/link (_agents_init_find $pf -type l -print | string join ,)
+check "find: no root -> status 1" 1 (_agents_init_find ""; echo $status)
+
 cleanup
 report