[BUG] Hookify plugin not passing blocking messages to Claude (missing permissionDecisionReason)

Fix confirmed working locally.

Added permissionDecisionReason to hookSpecificOutput in rule_engine.py:

return {
    "hookSpecificOutput": {
        "hookEventName": hook_event,
        "permissionDecision": "deny",
        "permissionDecisionReason": combined_message  # Added this line
    },
    "systemMessage": combined_message
}

Claude now receives the full detailed blocking message (tables, explanations, guidance) instead of just "Hook PreToolUse:Bash denied this tool".

Related: UserPromptSubmit hooks have the same problem

Just discovered that UserPromptSubmit hooks also don't pass messages to Claude when using systemMessage.

Root Cause

For UserPromptSubmit hooks, the message must be in hookSpecificOutput.additionalContext, not systemMessage.

Fix

# Before (broken)
return {
    "systemMessage": message
}

# After (working)
return {
    "hookSpecificOutput": {
        "hookEventName": "UserPromptSubmit",
        "additionalContext": message
    }
}

Summary of hook output fields needed for Claude to receive messages:

| Hook Event | Field Required |
|------------|----------------|
| PreToolUse | hookSpecificOutput.permissionDecisionReason |
| PostToolUse | hookSpecificOutput.permissionDecisionReason |
| UserPromptSubmit | hookSpecificOutput.additionalContext |
| Stop | reason (for blocking) |

This should probably be documented more clearly in the hooks documentation - systemMessage appears to only be shown to the user, while hook-specific fields are needed to pass context to Claude.

This issue has been inactive for 30 days. If the issue is still occurring, please comment to let us know. Otherwise, this issue will be automatically closed in 30 days for housekeeping purposes.

So both warn and block reason does not get back to Claude. This defeats the purpose of these pretty much.

This is now fixed in hookify-plus!

Added permissionDecisionReason to the hookSpecificOutput so Claude sees the detailed blocking message:

"hookSpecificOutput": {
    "hookEventName": hook_event,
    "permissionDecision": "deny",
    "permissionDecisionReason": combined_message  # Now Claude sees WHY!
}

# Install hookify-plus
ln -sf /path/to/hookify-plus ~/.claude/plugins/cache/claude-code-plugins/hookify/0.1.0

hookify-plus aggregates community fixes that Anthropic hasn't merged. Feel free to use it! 🚀

Feature was implemented in v2.1.9.
-Added support for PreToolUse hooks to return additionalContext to the model

👍

Confirmed: Still broken in hookify@claude-plugins-official (January 2026)

Just tested with Claude Code 2.1.11 and the official hookify plugin. The fix described above was never merged.

Test Setup

• Claude Code version: 2.1.11
• Hookify rule with enabled: true (native plugin handling)
• Rule: block base64 decode commands

Result

When the rule blocks a command, Claude receives only:

"Hook PreToolUse:Bash denied this tool"

Claude does NOT receive the detailed message explaining why it was blocked or what alternatives to use.

Proof - Current hookify source still missing the fix

Source: https://github.com/anthropics/claude-plugins-official/blob/main/plugins/hookify/core/rule_engine.py#L60-L79

# Line 62-63: combined_message contains the detailed rule message
messages = [f"**[{r.name}]**\n{r.message}" for r in blocking_rules]
combined_message = "\n\n".join(messages)

# Line 72-79: PreToolUse/PostToolUse response
return {
    "hookSpecificOutput": {
        "hookEventName": hook_event,
        "permissionDecision": "deny"
        # ❌ permissionDecisionReason is MISSING - Claude never sees combined_message!
    },
    "systemMessage": combined_message  # ✅ User sees this, but Claude doesn't
}

The combined_message has the detailed explanation, but it's only in systemMessage (shown to user). Claude needs it in permissionDecisionReason to understand why the command was blocked.

Workaround

We use a native hook bridge that outputs to stderr + exit 2, which Claude Code properly routes to the model. This works, but it's a workaround for what should be fixed in the plugin.

Request

Please add permissionDecisionReason to the hookSpecificOutput:

return {
    "hookSpecificOutput": {
        "hookEventName": hook_event,
        "permissionDecision": "deny",
        "permissionDecisionReason": combined_message  # ✅ Add this line
    },
    "systemMessage": combined_message
}

This is a one-line fix that would make hookify actually useful for guiding Claude's behavior.

This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.