claude-dev-env 8.98.11 → 8.99.1

package/.agents/agents/reference/pr-description-illustrative-voice.md

@@ -45,6 +45,6 @@ The person running the loop asked for a full look, so every pass gets a full loo
 - Prefer Before / After when the change is about how something is asked, found, checked, saved, retried, or stopped.
 - Use picture words: hand it, open it, find it, check it, save it, skip it, stop, call by name.
 - Prefer spoken results: “found it,” “missing,” “duplicated,” “couldn’t.”
-- Keep `Why` short and concrete.
-- Put proof only in `Verification`.
+- Keep `What this does` short and concrete.
+- Put proof only in `Proof in practice`.
 - Explain jargon immediately or replace it.

package/.agents/agents/reference/pr-description-verification.md

@@ -2,7 +2,7 @@
 
 Use this only when the PR writer needs a concrete verification sample.
 
-Verification should lead with something a reviewer can open, see, click, compare, or try.
+Proof in practice should lead with something a reviewer can open, see, click, compare, or try.
 
 Good:
 

package/.agents/agents/test_pr_description_writer_style.py

@@ -30,8 +30,8 @@ def test_voice_reference_keeps_the_plain_illustrative_shape() -> None:
         "found it",
         "missing",
         "duplicated",
-        "Why",
-        "Verification",
+        "What this does",
+        "Proof in practice",
     ):
         assert required_text in reference_text
 

package/.agents/skills/pull-request/SKILL.md

@@ -92,10 +92,9 @@ Check each behavior claim against the final diff and verification evidence.
 Preserve whether a rule is added, removed, or narrowed, and distinguish tests
 added from tests run. Refresh these claims after a rebase or correction.
 
-Open with the concrete problem and resulting behavior in one or two short
-paragraphs of plain prose. Write for a reviewer who has not read the worker's
-conversation. Include only claims supported by the final diff. State commands,
-results, and material limits under `## Verification`.
+Write the body in the section order `reference/body-sections.md` gives: What
+this does, What changed, Existing work, Proof in practice, Eval for a `feat`
+title, Risk, How to revert. Include only claims supported by the final diff.
 
 Keep substantive review and test evidence. Put extensive logs, raw diffs, token
 counts, and agent transcripts in an existing linked artifact or a closed
@@ -189,6 +188,7 @@ Read `reference/examples.md` for a create, a rejected comment, and a recovery.
 | `reference/publication-tasks.md` | Ordered tasks for every publication gate |
 | `reference/legacy-author-recovery.md` | Recovery of one selected legacy author record |
 | `reference/examples.md` | Worked create, rejected comment, and recovery examples |
+| `reference/body-sections.md` | Pull request body sections, their order, and what each holds |
 | `scripts/pull_request.py` | Command for author selection and create, edit, comment, and review actions |
 | `scripts/github_pr_command_constants/` | Constants for pull request actions, recovery, exit codes, and generic messages |
 | `scripts/test_pull_request.py` | Tests for `pull_request.py` |

package/.agents/skills/pull-request/reference/body-sections.md

@@ -0,0 +1,17 @@
+# Pull request body sections
+
+A reviewer reads the body in place of the code. The body says what the change does, how it was tested, what could break, and how to undo it. The order follows the Claude Code team's [code review article](https://claude.dev/blog/solving-code-review-at-anthropic/), whose template "asks how the change was tested and how to undo it". The opening sections follow the team's public pull requests, such as [claude-code #69470](https://github.com/anthropics/claude-code/pull/69470), which opens with What this does, Why now and What changed.
+
+| Section | Holds |
+|---|---|
+| `## What this does` | Two or three plain sentences: the behavior before, the behavior after, and who notices. Written for a reviewer who has not read the worker's conversation. |
+| `## What changed` | One bullet per area of the diff, in plain words. |
+| `## Existing work` | What already did part of the job, with a link or file and line, and what this change adds. Or the places searched. |
+| `## Proof in practice` | Each command in backticks, the output without the change and with it, and anything still unproven. |
+| `## Eval` | Only for a `feat` title. The eval's cases and grader, the command, and its result. |
+| `## Risk` | What could break, for whom, and how anyone would notice. `Low` and one reason for a doc or test change. |
+| `## How to revert` | The exact undo step, and any data or state a plain revert leaves behind. |
+
+A repository's own sections, such as a request record or a session link, follow these. Optional `## Why now` and `## After merge` sections go after What this does and after How to revert.
+
+`pull_request.py template` writes the headings in this order into a new body, and appends each missing heading to the end of a body that already has some. `durable_post_lint.py` reports a create body missing What this does, Risk, or How to revert, and the proof gate holds one missing Existing work, Proof in practice, or Eval for a `feat` title.

package/.agents/skills/pull-request/scripts/github_pr_command_constants/config/constants.py

@@ -9,11 +9,14 @@ ACTION_REVIEW = "review"
 ACTION_TEMPLATE = "template"
 
 ALL_BODY_TEMPLATE_HEADINGS: tuple[str, ...] = (
-    "Why",
+    "What this does",
+    "What changed",
     "Existing work",
     "Proof in practice",
-    "Verification",
+    "Risk",
+    "How to revert",
 )
+FEATURE_BODY_TEMPLATE_HEADING_FOLLOWS = "Proof in practice"
 FEATURE_BODY_TEMPLATE_HEADING = "Eval"
 FEATURE_TITLE_PATTERN: re.Pattern[str] = re.compile(r"^\s*feat(\([^)]*\))?!?:")
 BODY_TEMPLATE_HEADING_FORMAT = "## {heading}"

package/.agents/skills/pull-request/scripts/pull_request.py

@@ -26,6 +26,7 @@ from github_pr_command_constants.config.constants import (
     BODY_TEMPLATE_HEADING_FORMAT,
     BODY_TEMPLATE_LINE_END,
     FEATURE_BODY_TEMPLATE_HEADING,
+    FEATURE_BODY_TEMPLATE_HEADING_FOLLOWS,
     FEATURE_TITLE_PATTERN,
     ALL_GITHUB_AUTHORIZATION_ENVIRONMENT_KEYS,
     ALL_LINTER_ACTIONS_BY_COMMAND,
@@ -77,13 +78,11 @@ def _build_parser() -> argparse.ArgumentParser:
 
 
 def _template_headings(title: str) -> list[str]:
-    all_headings = [
-        BODY_TEMPLATE_HEADING_FORMAT.format(heading=each_heading)
-        for each_heading in ALL_BODY_TEMPLATE_HEADINGS
-    ]
+    all_heading_names = list(ALL_BODY_TEMPLATE_HEADINGS)
     if FEATURE_TITLE_PATTERN.match(title):
-        all_headings.append(BODY_TEMPLATE_HEADING_FORMAT.format(heading=FEATURE_BODY_TEMPLATE_HEADING))
-    return all_headings
+        eval_position = all_heading_names.index(FEATURE_BODY_TEMPLATE_HEADING_FOLLOWS) + 1
+        all_heading_names.insert(eval_position, FEATURE_BODY_TEMPLATE_HEADING)
+    return [BODY_TEMPLATE_HEADING_FORMAT.format(heading=each_heading) for each_heading in all_heading_names]
 
 
 def write_body_template(body_file: Path, title: str) -> list[str]:
@@ -91,8 +90,9 @@ def write_body_template(body_file: Path, title: str) -> list[str]:
 
     ::
 
-        new file, title "fix: x"   -> ["## Why", "## Existing work", "## Proof in practice", "## Verification"]
-        new file, title "feat: x"  -> the same plus "## Eval"
+        new file, title "fix: x"   -> ["## What this does", "## What changed", "## Existing work",
+                                       "## Proof in practice", "## Risk", "## How to revert"]
+        new file, title "feat: x"  -> the same with "## Eval" after "## Proof in practice"
         file already has them all  -> [] and the file is unchanged
 
     Args:

package/.agents/skills/pull-request/scripts/test_pull_request.py

@@ -544,3 +544,20 @@ def should_keep_written_text_and_add_only_missing_headings(tmp_path: Path) -> No
     assert first_text.startswith("## Why\nThe gate bounced twice.\n")
     assert first_text.count("## Why") == 1
     assert body_path.read_text(encoding="utf-8") == first_text
+
+
+def should_write_the_reader_headings_in_order_with_eval_after_the_proof(tmp_path: Path) -> None:
+    body_path = tmp_path / BODY_FILENAME
+    _run_template(body_path, "feat(hooks): add the gate")
+    all_heading_lines = [
+        each_line for each_line in body_path.read_text(encoding="utf-8").splitlines() if each_line.startswith("## ")
+    ]
+    assert all_heading_lines == [
+        "## What this does",
+        "## What changed",
+        "## Existing work",
+        "## Proof in practice",
+        "## Eval",
+        "## Risk",
+        "## How to revert",
+    ]

package/bin/install-pstack-plugin.mjs

@@ -91,12 +91,13 @@ const HOST_DEFINITIONS = Object.freeze({
  * verb gets refresh commands: the marketplace add and plugin install leave an
  * existing entry in place, and the update commands then pull the newest
  * published version over it, so an installed plugin stays installed through
- * the whole run. When any refresh command fails, or the host has no update
- * verb, the removal commands run and take out each installed plugin, its
- * cached versions, and the marketplace clone, and the install commands that
- * follow fetch the newest published version into an empty slot. Removing a
- * marketplace uninstalls each plugin it supplied, so the plan removes and adds
- * that marketplace once for the whole set.
+ * the whole run. When a refresh command fails for a reason only a rebuild
+ * fixes, or the host has no update verb, the removal commands run and take
+ * out each installed plugin, its cached versions, and the marketplace clone,
+ * and the install commands that follow fetch the newest published version
+ * into an empty slot. Removing a marketplace uninstalls each plugin it
+ * supplied, so the plan removes and adds that marketplace once for the whole
+ * set.
  *
  * Claude Code and Codex publish a marketplace under different plugin verbs, so
  * the install verb belongs to the host and the marketplace and identifier
@@ -219,7 +220,12 @@ function runHostCommand(executable, commandArguments, options) {
         env: { ...process.env, ...options.environment },
         windowsVerbatimArguments: invocation.windowsVerbatimArguments,
     });
-    return { status: spawned.status, stderr: spawned.stderr ?? '', error: spawned.error };
+    return {
+        status: spawned.status,
+        stdout: spawned.stdout ?? '',
+        stderr: spawned.stderr ?? '',
+        error: spawned.error,
+    };
 }
 
 const COMMAND_NOT_FOUND_EXIT_CODE = 9009;
@@ -297,15 +303,33 @@ function lockedFolderDetail(allSpecs, host, homeDirectory, errorText, findLocker
     }).join('');
 }
 
+const REBUILD_REQUIRED_PATTERN = /doesn't match its extraKnownMarketplaces entry|Source path does not exist/;
+
+function commandFailureDetail(outcome) {
+    return singleLineDetail(outcome.stderr) || singleLineDetail(outcome.stdout)
+        || outcome.error?.message || `exit ${outcome.status}`;
+}
+
+function failedCommandOutcome(allSpecs, host, homeDirectory, executable, commandArguments, outcome, findLockers) {
+    const detail = commandFailureDetail(outcome);
+    const lockAdvice = lockedFolderDetail(allSpecs, host, homeDirectory, detail, findLockers);
+    return {
+        host,
+        executable,
+        status: 'failed',
+        warning: `${executable} ${commandArguments.join(' ')} failed: ${detail}${lockAdvice}`,
+    };
+}
+
 function refreshInPlace(plan, executable, commandEnvironment, runCommand) {
     for (const commandArguments of plan.refreshCommands) {
         const outcome = runCommand(executable, [...commandArguments], {
             environment: commandEnvironment,
         });
-        if (namesAnAbsentCommand(outcome)) return 'absent';
-        if (outcome.status !== 0 || outcome.error) return 'failed';
+        if (namesAnAbsentCommand(outcome)) return { state: 'absent' };
+        if (outcome.status !== 0 || outcome.error) return { state: 'failed', commandArguments, outcome };
     }
-    return 'refreshed';
+    return { state: 'refreshed' };
 }
 
 function removeOrphanedVersionsForHost(allSpecs, host, homeDirectory) {
@@ -326,11 +350,18 @@ function installForHost(allSpecs, host, homeDirectory, environment, runCommand,
     }
     if (plan.refreshCommands.length > 0) {
         const refreshOutcome = refreshInPlace(plan, executable, commandEnvironment, runCommand);
-        if (refreshOutcome === 'absent') return absentHostOutcome(allSpecs, host, executable);
-        if (refreshOutcome === 'refreshed') {
+        if (refreshOutcome.state === 'absent') return absentHostOutcome(allSpecs, host, executable);
+        if (refreshOutcome.state === 'refreshed') {
             removeOrphanedVersionsForHost(allSpecs, host, homeDirectory);
             return { host, executable, status: 'installed', warning: null };
         }
+        const refreshOutput = `${refreshOutcome.outcome.stderr ?? ''}\n${refreshOutcome.outcome.stdout ?? ''}`;
+        if (!REBUILD_REQUIRED_PATTERN.test(refreshOutput)) {
+            return failedCommandOutcome(
+                allSpecs, host, homeDirectory, executable,
+                refreshOutcome.commandArguments, refreshOutcome.outcome, findLockers,
+            );
+        }
     }
     for (const commandArguments of plan.removalCommands) {
         const outcome = runCommand(executable, [...commandArguments], {
@@ -347,14 +378,9 @@ function installForHost(allSpecs, host, homeDirectory, environment, runCommand,
         });
         if (namesAnAbsentCommand(outcome)) return absentHostOutcome(allSpecs, host, executable);
         if (outcome.status !== 0 || outcome.error) {
-            const detail = singleLineDetail(outcome.stderr) || outcome.error?.message || `exit ${outcome.status}`;
-            const lockAdvice = lockedFolderDetail(allSpecs, host, homeDirectory, detail, findLockers);
-            return {
-                host,
-                executable,
-                status: 'failed',
-                warning: `${executable} ${commandArguments.join(' ')} failed: ${detail}${lockAdvice}`,
-            };
+            return failedCommandOutcome(
+                allSpecs, host, homeDirectory, executable, commandArguments, outcome, findLockers,
+            );
         }
     }
     removeOrphanedVersionsForHost(allSpecs, host, homeDirectory);
@@ -385,12 +411,14 @@ export function restartNotice(allSpecs, allInstalledHosts) {
  * plugin, both of which leave an existing entry alone, then updates the
  * marketplace and each plugin. An installed plugin stays installed and enabled
  * through the run, so a session that starts mid-run still loads it, and a
- * failed network step leaves the prior version in place. When a refresh
- * command fails, such as a marketplace add whose sparse paths changed, the
- * host falls back to the clean install Codex always runs: it removes every
- * plugin in the set and then the marketplace, so an older version never
- * survives beside the new one, and adds them again. A removal that finds
- * nothing to remove exits non-zero on a fresh home, and the install goes on.
+ * failed refresh command reports the host failed and leaves the prior version
+ * in place. Two refresh errors need a rebuild: a source that no longer
+ * matches the settings entry, and a source path the sparse clone lacks. The
+ * host reads either from the failing refresh command's output, then falls
+ * back to the clean install Codex always runs: it removes every plugin in the
+ * set and then the marketplace, so an older version never survives beside the
+ * new one, and adds them again. A removal that finds nothing to remove exits
+ * non-zero on a fresh home, and the install goes on.
  * Before each Claude marketplace add, the install deletes the half-made clone
  * folder an interrupted earlier add left, so the add starts in an empty slot.
  *
@@ -482,7 +510,7 @@ export function removeRetiredMarketplacePlugins(allRetiredSpecs, options, depend
         );
         if (namesAnAbsentCommand(outcome)) break;
         if (outcome.status !== 0 || outcome.error) {
-            const detail = singleLineDetail(outcome.stderr) || outcome.error?.message || `exit ${outcome.status}`;
+            const detail = commandFailureDetail(outcome);
             failedRemovals.push({ pluginIdentifier: eachSpec.pluginIdentifier, detail });
             continue;
         }

package/bin/install-pstack-plugin.test.mjs

@@ -117,6 +117,30 @@ test('a failed Claude refresh falls back to removing and adding the marketplace
     ]);
 });
 
+test('a Claude refresh error printed on standard output still triggers the rebuild', () => {
+    const runner = recordingRunner((executable, commandArguments) => (
+        commandArguments[2] === 'add' && runner.calls.length === 1
+            ? { status: 1, stdout: "its source doesn't match its extraKnownMarketplaces entry", stderr: '' }
+            : { status: 0, stderr: '' }));
+    const outcome = installMarketplacePlugins([USAGE_WRAPUP_PLUGIN_SPEC], { ...ROOTS, hosts: ['claude'] }, runner);
+    assert.equal(outcome.status, 'installed');
+    assert.ok(runner.calls.some(call => call.commandArguments.includes('uninstall')));
+});
+
+test('a Claude refresh that fails on the network keeps the installed plugin and reports the failure', () => {
+    const runner = recordingRunner((executable, commandArguments) => (
+        commandArguments[1] === 'update'
+            ? { status: 1, stderr: 'Failed to update plugin: getaddrinfo ENOTFOUND github.com' }
+            : { status: 0, stderr: '' }));
+    const outcome = installMarketplacePlugins([PSTACK_PLUGIN_SPEC], { ...ROOTS, hosts: ['claude'] }, runner);
+    assert.equal(outcome.status, 'failed');
+    assert.match(outcome.hosts[0].warning, /claude plugin update pstack@pstack-claude failed: .*ENOTFOUND/);
+    const allRemovalCalls = runner.calls.filter(call => (
+        call.commandArguments.includes('uninstall') || call.commandArguments.includes('remove')
+    ));
+    assert.deepEqual(allRemovalCalls, []);
+});
+
 test('each host removes the installed plugin and marketplace before a clean install', () => {
     assert.deepEqual(marketplacePluginsPlan([PSTACK_PLUGIN_SPEC], 'claude').removalCommands, [
         ['plugin', 'uninstall', PSTACK_PLUGIN_IDENTIFIER],
@@ -131,9 +155,10 @@ test('each host removes the installed plugin and marketplace before a clean inst
 test('a removal that finds nothing installed still lets the install run', () => {
     const runner = recordingRunner((executable, commandArguments) => (
         commandArguments.includes('uninstall') || commandArguments.includes('remove')
-            || commandArguments.includes('update')
             ? { status: 1, stderr: 'not installed' }
-            : { status: 0, stderr: '' }));
+            : commandArguments.includes('update')
+                ? { status: 1, stderr: 'Source path does not exist: plugins/pstack' }
+                : { status: 0, stderr: '' }));
     const outcome = installMarketplacePlugins([PSTACK_PLUGIN_SPEC], { ...ROOTS, hosts: ['claude'] }, runner);
     assert.equal(outcome.status, 'installed');
     assert.deepEqual(runner.calls.at(-1).commandArguments, ['plugin', 'install', PSTACK_PLUGIN_IDENTIFIER]);

package/bin/install.settings-defaults.test.mjs

@@ -35,6 +35,11 @@ const INSTALL_ENTRY = join(PACKAGE_ROOT, 'bin', 'install.mjs');
 const EXPECTED_DENY_ENTRIES = [];
 const EXPECTED_ALLOW_ENTRIES = [
     'WebFetch(domain:docs.github.com)',
+    'WebFetch(domain:claude.com)',
+    'WebFetch(domain:*.claude.com)',
+    'WebFetch(domain:claude.dev)',
+    'WebFetch(domain:*.claude.dev)',
+    'WebFetch(domain:www.anthropic.com)',
     'Read(~/.claude/docs/**)',
 ];
 const PROFILE_ALLOW_ENTRIES = DECLARED_PROFILE_SETTINGS
@@ -76,7 +81,7 @@ function runInstallerInSandbox(sandboxHome, installerArguments = []) {
     });
 }
 
-test('package settings.json publishes the GitHub docs fetch allow and no managed deny entries', () => {
+test('package settings.json publishes the docs fetch allows and no managed deny entries', () => {
     assert.deepEqual(packageManagedPermissions(), {
         allow: EXPECTED_ALLOW_ENTRIES,
         deny: EXPECTED_DENY_ENTRIES,

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "claude-dev-env",
-    "version": "8.98.11",
+    "version": "8.99.1",
     "description": "Claude Code development standards — rules, hooks, agents, commands, and skills",
     "type": "module",
     "bin": {

package/scripts/durable_post_lint.py

@@ -25,7 +25,6 @@ from durable_post_lint_config.config.constants import (
     ALL_BODY_REQUIRED_ACTIONS,
     ALL_PATH_ANCHORED_VOLATILE_PATH_MARKERS,
     ALL_POST_ACTIONS,
-    ALL_PR_DESCRIPTION_ACTIONS,
     ALL_RELEASE_BODY_MARKERS,
     ALL_REQUIRED_PR_DESCRIPTION_HEADINGS,
     ALL_TITLE_ACTIONS,
@@ -266,9 +265,8 @@ def _body_findings(
     if _is_release_automation_branch(head_branch):
         all_findings.extend(_release_body_findings(body_text))
     else:
-        if action in ALL_PR_DESCRIPTION_ACTIONS:
-            all_findings.extend(_description_heading_findings(body_text))
         if action == ACTION_PR_CREATE:
+            all_findings.extend(_description_heading_findings(body_text))
             all_findings.extend(_pull_request_proof_findings(title, body_text))
         if action == ACTION_GITHUB_MCP_COMMENT:
             all_findings.extend(_comment_length_findings(body_text))

package/scripts/durable_post_lint_config/config/constants.py

@@ -37,9 +37,6 @@ ALL_BODY_REQUIRED_ACTIONS: frozenset[str] = frozenset(
         ACTION_PR_REVIEW,
     }
 )
-ALL_PR_DESCRIPTION_ACTIONS: frozenset[str] = frozenset(
-    {ACTION_PR_CREATE, ACTION_PR_EDIT}
-)
 ALL_TITLE_ACTIONS: frozenset[str] = frozenset({ACTION_PR_CREATE, ACTION_PR_EDIT})
 ALL_CONVENTIONAL_TITLE_TYPES: tuple[str, ...] = (
     "build",
@@ -58,8 +55,9 @@ CONVENTIONAL_TITLE_PATTERN: re.Pattern[str] = re.compile(
     rf"^(?:{'|'.join(ALL_CONVENTIONAL_TITLE_TYPES)})(?:\([^)\r\n]+\))?!?: [^\s].*$"
 )
 ALL_REQUIRED_PR_DESCRIPTION_HEADINGS: tuple[str, ...] = (
-    "Why",
-    "Verification",
+    "What this does",
+    "Risk",
+    "How to revert",
 )
 ALL_PATH_ANCHORED_VOLATILE_PATH_MARKERS: tuple[str, ...] = (
     ".claude-profile-a/jobs/",

package/scripts/durable_post_lint_config/config/test_constants.py

@@ -40,5 +40,5 @@ def should_name_the_branch_prefix_release_automation_pushes() -> None:
     assert RELEASE_BRANCH_PREFIX == "release-please--branches--"
 
 
-def should_require_only_the_headings_both_pull_request_rule_sets_name() -> None:
-    assert ALL_REQUIRED_PR_DESCRIPTION_HEADINGS == ("Why", "Verification")
+def should_require_the_reader_headings_a_reviewer_reads_in_place_of_the_code() -> None:
+    assert ALL_REQUIRED_PR_DESCRIPTION_HEADINGS == ("What this does", "Risk", "How to revert")

package/scripts/test_durable_post_lint.py

@@ -18,9 +18,9 @@ from hooks_constants.pull_request_proof_constants import (
 )
 
 SCRIPT_PATH = Path(__file__).resolve().parent / "durable_post_lint.py"
-VALID_PR_BODY = """## Why
+VALID_PR_BODY = """## What this does
 
-Why text.
+What this does text.
 
 ## Scope
 
@@ -34,9 +34,13 @@ Tradeoffs text.
 
 Blast radius text.
 
-## Verification
+## Risk
 
-Verification text.
+Risk text.
+
+## How to revert
+
+Revert text.
 
 ## Existing work
 
@@ -201,7 +205,7 @@ def test_pr_body_reports_each_missing_heading(missing_heading: str) -> None:
     ]
 
 
-def should_accept_an_edit_body_carrying_only_the_two_required_headings() -> None:
+def should_accept_an_edit_body_without_the_reader_headings() -> None:
     all_findings = durable_post_lint.lint_durable_post(
         action="pr-edit",
         title="fix(cli): tighten input handling",

package/settings.json

@@ -3,6 +3,11 @@
   "permissions": {
     "allow": [
       "WebFetch(domain:docs.github.com)",
+      "WebFetch(domain:claude.com)",
+      "WebFetch(domain:*.claude.com)",
+      "WebFetch(domain:claude.dev)",
+      "WebFetch(domain:*.claude.dev)",
+      "WebFetch(domain:www.anthropic.com)",
       "Read(~/.claude/docs/**)"
     ]
   },