diff --git a/README.md b/README.md
index 179d81ca..a6030772 100644
--- a/README.md
+++ b/README.md
@@ -140,6 +140,7 @@ Even only with following subscriptions, ultrawork will work well (this project i
 | :---: | :--------------------------- | :---------------------------------------------------------------------------------------------------------------------------------- |
 |   🤖   | **Discipline Agents**        | Sisyphus orchestrates Hephaestus, Oracle, Librarian, Explore. A full AI dev team in parallel.                                       |
 |   ⚡   | **`ultrawork` / `ulw`**      | One word. Every agent activates. Doesn't stop until done.                                                                           |
+|   🚪   | **IntentGate**                 | Analyzes true user intent before classifying or acting. No more literal misinterpretations.                                         |
 |   🔗   | **Hash-Anchored Edit Tool**  | `LINE#ID` content hash validates every change. Zero stale-line errors.                                                              |
 |   🛠️   | **LSP + AST-Grep**           | Workspace rename, pre-build diagnostics, AST-aware rewrites. IDE precision for agents.                                              |
 |   🧠   | **Background Agents**        | Fire 5+ specialists in parallel. Context stays lean. Results when ready.                                                            |
diff --git a/docs/features.md b/docs/features.md
index 7265fb87..57a86361 100644
--- a/docs/features.md
+++ b/docs/features.md
@@ -97,6 +97,19 @@ Customize agent models, prompts, and permissions in `oh-my-opencode.json`. See [
 
 ---
 
+## IntentGate
+
+Every prompt a user enters goes through the **IntentGate** before action is taken.
+
+Agent models often take user instructions too literally—resulting in premature execution or shallow responses. IntentGate fixes this.
+
+1. **Verbalization Before Action:** The orchestrator agent (Sisyphus) and the autonomous worker (Hephaestus) explicitly verbalize the user's *true* intent (e.g. Investigation, Implementation, Fix, Research) before proceeding.
+2. **Pre-Planning Analysis:** The `metis` plan consultant specifically looks for hidden intents, missing context, and ambiguity.
+3. **Exploration Default:** If the true intent implies missing knowledge, the agents will automatically spawn `explore` or `librarian` background tasks *before* taking action or asking clarification questions.
+
+IntentGate ensures your tasks are solved thoroughly rather than treated as simple string-matching instructions.
+
+---
 ## Skills: Specialized Knowledge
 
 Skills provide specialized workflows with embedded MCP servers and detailed instructions.