{"version":"2.1.323","anchor":"prompt-injection-hook-message-varies-by-tool","canonical_anchor":"prompt-injection-hook-message-varies-by-tool","heading":"Hook word-replacement message now depends on the tool","tier":"internal","area":"Hooks","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.323\/e\/prompt-injection-hook-message-varies-by-tool","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.323","markdown":"### Hook word-replacement message now depends on the tool\n\nWhen hook words are replaced, the message Claude Code appends now varies by tool instead of choosing between two\n\n**What**\n\nWhen Claude Code replaces certain words coming from a hook (a user command that runs at set points), it adds a message to the output. Previously that message was one of two. It now depends on the tool involved:\n\n- One specific tool gets its own message.\n\n- The plan-approval tool gets its own message.\n\n- Tools in a particular group get another message.\n\n- All other tools get a default message.\n\n**Why**\n\nThe model gets guidance text that is more specific to the tool in use.\n\n- Area: Hooks\n- Tier: Under the hood\n- Useful: 1\/5\n- Signal: 1\/5\n- Scope: individual\n- Heads-up: no"}