* docs: add investigation reports for 5 open GitHub issues Comprehensive analysis of issues #543, #544, #545, #555, and #557: - #557: settings.json not generated, module loader error (node/bun mismatch) - #555: Windows hooks not executing, hasIpc always false - #545: formatTool crashes on non-JSON tool_input strings - #544: mem-search skill hint shown incorrectly to Claude Code users - #543: /claude-mem slash command unavailable despite installation Each report includes root cause analysis, affected files, and proposed fixes. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * fix(logger): handle non-JSON tool_input in formatTool (#545) Wrap JSON.parse in try-catch to handle raw string inputs (e.g., Bash commands) that aren't valid JSON. Falls back to using the string as-is. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * fix(context): update mem-search hint to reference MCP tools (#544) Update hint messages to reference MCP tools (search, get_observations) instead of the deprecated "mem-search skill" terminology. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * fix(settings): auto-create settings.json on first load (#557, #543) When settings.json doesn't exist, create it with defaults instead of returning in-memory defaults. Creates parent directory if needed. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * fix(hooks): use bun runtime for hooks except smart-install (#557) Change hook commands from node to bun since hooks use bun:sqlite. Keep smart-install.js on node since it bootstraps bun installation. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * chore: rebuild plugin scripts * docs: clarify that build artifacts must be committed * fix(docs): update build artifacts directory reference in CLAUDE.md * test: add test coverage for PR #558 fixes - Fix 2 failing tests: update "mem-search skill" → "MCP tools" expectations - Add 56 tests for formatTool() JSON.parse crash fix (Issue #545) - Add 27 tests for settings.json auto-creation (Issue #543) Test coverage includes: - formatTool: JSON parsing, raw strings, objects, null/undefined, all tool types - Settings: file creation, directory creation, schema migration, edge cases 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * fix(tests): clean up flaky tests and fix circular dependency Phase 1 of test quality improvements: - Delete 6 harmful/worthless test files that used problematic mock.module() patterns or tested implementation details rather than behavior: - context-builder.test.ts (tested internal implementation) - export-types.test.ts (fragile mock patterns) - smart-install.test.ts (shell script testing antipattern) - session_id_refactor.test.ts (outdated, tested refactoring itself) - validate_sql_update.test.ts (one-time migration validation) - observation-broadcaster.test.ts (excessive mocking) - Fix circular dependency between logger.ts and SettingsDefaultsManager.ts by using late binding pattern - logger now lazily loads settings - Refactor mock.module() to spyOn() in several test files for more maintainable and less brittle tests: - observation-compiler.test.ts - gemini_agent.test.ts - error-handler.test.ts - server.test.ts - response-processor.test.ts All 649 tests pass. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * refactor(tests): phase 2 - reduce mock-heavy tests and improve focus - Remove mock-heavy query tests from observation-compiler.test.ts, keep real buildTimeline tests - Convert session_id_usage_validation.test.ts from 477 to 178 lines of focused smoke tests - Remove tests for language built-ins from worker-spawn.test.ts (JSON.parse, array indexing) - Rename logger-coverage.test.ts to logger-usage-standards.test.ts for clarity 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * docs(tests): phase 3 - add JSDoc mock justification to test files Document mock usage rationale in 5 test files to improve maintainability: - error-handler.test.ts: Express req/res mocks, logger spies (~11%) - fallback-error-handler.test.ts: Zero mocks, pure function tests - session-cleanup-helper.test.ts: Session fixtures, worker mocks (~19%) - hook-constants.test.ts: process.platform mock for Windows tests (~12%) - session_store.test.ts: Zero mocks, real SQLite :memory: database Part of ongoing effort to document mock justifications per TESTING.md guidelines. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * test(integration): phase 5 - add 72 tests for critical coverage gaps Add comprehensive test coverage for previously untested areas: - tests/integration/hook-execution-e2e.test.ts (10 tests) Tests lifecycle hooks execution flow and context propagation - tests/integration/worker-api-endpoints.test.ts (19 tests) Tests all worker service HTTP endpoints without heavy mocking - tests/integration/chroma-vector-sync.test.ts (16 tests) Tests vector embedding synchronization with ChromaDB - tests/utils/tag-stripping.test.ts (27 tests) Tests privacy tag stripping utilities for both <private> and <meta-observation> tags All tests use real implementations where feasible, following the project's testing philosophy of preferring integration-style tests over unit tests with extensive mocking. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * context update * docs: add comment linking DEFAULT_DATA_DIR locations Added NOTE comment in logger.ts pointing to the canonical DEFAULT_DATA_DIR in SettingsDefaultsManager.ts. This addresses PR reviewer feedback about the fragility of having the default defined in two places to avoid circular dependencies. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
8.8 KiB
Issue #555 Analysis: Windows Hooks Not Executing - hasIpc Always False
Date: 2026-01-05 Version Analyzed: 8.5.9 Claude Code Version: 2.0.76 Platform: Windows 11 (Build 26100), Git Bash (MINGW64) Status: INVESTIGATION COMPLETE - Root cause identified
Issue Summary
On Windows 11 with Git Bash, Claude-mem plugin hooks are not executing at all. While the worker service starts successfully and responds to health checks, no observations are being saved and no hook-related logs appear.
Reported Symptoms
// /api/health
{
"status": "ok",
"build": "TEST-008-wrapper-ipc",
"managed": false,
"hasIpc": false,
"platform": "win32",
"pid": 3596,
"initialized": true,
"mcpReady": true
}
// /api/stats
{
"observations": 0,
"sessions": 1
}
Key Observations
- Worker starts and responds correctly to HTTP requests
hasIpcisfalse(this is expected behavior, not a bug)observationsremains at0- no data being captured- No
[HOOK]entries in worker logs - hooks never execute - This differs from issue #517 which was about PowerShell escaping
Root Cause Analysis
Primary Cause: Hook Commands Not Executing
The hooks defined in plugin/hooks/hooks.json are never being invoked by Claude Code on Windows.
Understanding hasIpc
The hasIpc field is a red herring and is working as intended:
// src/services/server/Server.ts:152
hasIpc: typeof process.send === 'function'
This checks if the worker process was spawned with an IPC channel (via fork() or spawn() with stdio: 'ipc'). Plugin hooks execute as independent command-line processes, NOT as forked child processes with IPC channels. Therefore, hasIpc: false is the expected, normal behavior for all hook executions.
Actual Problem: Hook Command Execution Failure
The hooks.json uses Unix-style environment variable syntax:
{
"command": "bun \"${CLAUDE_PLUGIN_ROOT}/scripts/worker-service.cjs\" start"
}
On Windows, this fails because:
-
Shell Interpreter Mismatch: Claude Code on Windows likely uses
cmd.exeor PowerShell to execute hook commands, not Git Bash. The${VARIABLE}syntax only works in Bash; cmd.exe uses%VARIABLE%. -
PATH Environment Differences: The user runs Claude in Git Bash where
bunandnodeare in PATH. However, Claude Code executes hooks in its own shell context (likely cmd.exe), which may not inherit Git Bash's PATH configuration. -
CLAUDE_PLUGIN_ROOT Resolution: If Claude Code doesn't properly set or expand
CLAUDE_PLUGIN_ROOTbefore executing the command, the entire path becomes invalid.
Code Investigation Findings
Affected Files
| File | Purpose | Issue |
|---|---|---|
plugin/hooks/hooks.json |
Hook command definitions | Uses ${CLAUDE_PLUGIN_ROOT} Unix syntax |
plugin/scripts/smart-install.js |
Dependency installer | Executed via hooks.json, never runs on Windows |
plugin/scripts/worker-service.cjs |
Worker CLI | Executed via hooks.json, never runs on Windows |
plugin/scripts/*.js |
Hook scripts | None execute because hooks.json commands fail |
hooks.json Analysis
Current hooks.json commands:
{
"SessionStart": [{
"hooks": [
{ "command": "node \"${CLAUDE_PLUGIN_ROOT}/scripts/smart-install.js\"" },
{ "command": "bun \"${CLAUDE_PLUGIN_ROOT}/scripts/worker-service.cjs\" start" },
{ "command": "node \"${CLAUDE_PLUGIN_ROOT}/scripts/context-hook.js\"" }
]
}],
"PostToolUse": [{
"hooks": [
{ "command": "bun \"${CLAUDE_PLUGIN_ROOT}/scripts/worker-service.cjs\" start" },
{ "command": "node \"${CLAUDE_PLUGIN_ROOT}/scripts/save-hook.js\"" }
]
}]
}
Problems identified:
${CLAUDE_PLUGIN_ROOT}- Unix variable expansion, fails in cmd.exebuncommand - May not be in system PATH on Windowsnodecommand - May not be in system PATH accessible to Claude Code
Worker hasIpc Usage
The hasIpc field is used only for admin endpoint IPC messaging, which is a separate concern from hook execution:
// src/services/server/Server.ts:209-216
const isWindowsManaged = process.platform === 'win32' &&
process.env.CLAUDE_MEM_MANAGED === 'true' &&
process.send;
if (isWindowsManaged) {
process.send!({ type: 'restart' });
}
This IPC mechanism is for managed process scenarios and is unrelated to why hooks aren't executing.
Relationship to Issue #517
| Aspect | Issue #517 | Issue #555 |
|---|---|---|
| Problem | PowerShell $_ variable misinterpreted by Bash |
Hooks not executing at all |
| Location | ProcessManager.ts (worker internals) | hooks.json execution by Claude Code |
| Fix Applied | Replaced PowerShell with WMIC | N/A (new issue) |
| Scope | Worker process management | Claude Code hook invocation |
Issue #517 fixed internal worker operations (orphaned process cleanup). Issue #555 is a completely different layer - it's about Claude Code's plugin system failing to invoke hooks on Windows.
Proposed Fix
Option 1: Cross-Platform Wrapper Script (Recommended)
Create a platform-aware wrapper that handles path resolution:
// plugin/scripts/hook-runner.js
#!/usr/bin/env node
const path = require('path');
const { spawn } = require('child_process');
// Resolve CLAUDE_PLUGIN_ROOT or compute from script location
const pluginRoot = process.env.CLAUDE_PLUGIN_ROOT ||
path.dirname(__dirname);
const hookScript = process.argv[2];
const hookPath = path.join(pluginRoot, 'scripts', hookScript);
// Execute the actual hook
require(hookPath);
Update hooks.json to use relative paths:
{
"command": "node ./scripts/hook-runner.js context-hook.js"
}
Option 2: Windows-Specific hooks.json
Create a Windows-compatible version using %CLAUDE_PLUGIN_ROOT% syntax:
{
"command": "node \"%CLAUDE_PLUGIN_ROOT%\\scripts\\smart-install.js\""
}
Drawback: Requires maintaining two hooks.json versions or using conditional logic.
Option 3: Use Absolute Paths
Generate hooks.json at install time with resolved absolute paths:
{
"command": "node \"C:\\Users\\username\\.claude\\plugins\\marketplaces\\thedotmack\\plugin\\scripts\\smart-install.js\""
}
Drawback: Less portable, requires install-time generation.
Option 4: Ensure bun/node in System PATH
Add installation validation to ensure bun and node are in the system-wide PATH, not just Git Bash's PATH:
# In smart-install.js for Windows
if (IS_WINDOWS) {
// Add to system PATH if not present
// Or use absolute paths to node/bun executables
}
Debugging Steps for Users
-
Verify plugin registration:
claude /status -
Check plugin installation:
dir $env:USERPROFILE\.claude\plugins\marketplaces\thedotmack\plugin\hooks -
Test environment variable:
$env:CLAUDE_PLUGIN_ROOT = "$env:USERPROFILE\.claude\plugins\marketplaces\thedotmack\plugin" node "$env:CLAUDE_PLUGIN_ROOT\scripts\smart-install.js" -
Check if node/bun are in system PATH:
where.exe node where.exe bun -
Enable Claude Code debug logging:
- Check Claude Code settings for debug/verbose mode
- Look for hook execution errors in logs
Impact Assessment
- Severity: High - Complete loss of memory functionality on Windows
- Scope: All Windows users, especially those using Git Bash
- Workaround: None currently - hooks must execute for memory to work
- Affected Versions: Likely affects 8.5.x on Windows with Claude Code 2.0.76+
Recommended Actions
- Immediate: Document the issue and potential workarounds
- Short-term: Implement Option 1 (cross-platform wrapper script)
- Long-term: Request clarification from Anthropic on Windows hook execution behavior
- Testing: Add Windows CI/CD testing for hook execution
Files to Modify
plugin/hooks/hooks.json- Update command syntaxplugin/scripts/hook-runner.js- New cross-platform wrapper (create)plugin/scripts/smart-install.js- Add PATH validation for Windowsdocs/public/troubleshooting.mdx- Document Windows hook issues
Appendix: Technical Details
Environment Variable Expansion by Shell
| Shell | Syntax | Works in hooks.json |
|---|---|---|
| Bash | ${VAR} or $VAR |
Yes (if Bash executes) |
| cmd.exe | %VAR% |
Yes (if cmd executes) |
| PowerShell | $env:VAR |
Yes (if PS executes) |
Claude Code Hook Execution Flow
- Claude Code loads hooks.json from plugin directory
- On hook event (SessionStart, PostToolUse, etc.), executes defined commands
- Commands are executed via system shell (platform-dependent)
- Hook process receives JSON via stdin, outputs response to stdout
- Claude Code processes hook output
The failure occurs at step 3 when the shell cannot resolve the command or environment variables.