Writing Hookify Rules

The skill enables the definition and dynamic loading of arbitrary 'Hookify Rules' from local files (`.claude/hookify.{rule-name}.local.md`). An attacker with write access to these files can inject persistent policies to gain broad control over the agent's behavior, including blocking or warning about specific actions (`action: block|warn`), monitoring various events (bash commands, file operations, user prompts), and potentially hijacking the agent's goals or performing reconnaissance.

Rules are stored in `.claude/hookify.{rule-name}.local.md` files. (Line 10)
rules are read dynamically on next tool use (Line 180)
`action`: `warn`|`block` (Lines 50-53)
`event`: `bash`|`file`|`stop`|`prompt` (Lines 40-45)
Condition fields: `command`, `file_path`, `new_text`, `old_text`, `content`, `user_prompt` (Lines 100-103, 230-232)

The `pattern` field uses Python regex syntax. Maliciously crafted complex regular expressions could lead to Regular Expression Denial of Service (ReDoS) when evaluated against input, potentially consuming excessive CPU resources and impacting agent performance or availability.

`pattern` (simple format): Regex pattern to match ... Python regex syntax (Lines 57-59)
operator: `regex_match` (Line 96)