diff --git a/app/server/README.md b/app/server/README.md
index e73ff30c..41f25086 100644
--- a/app/server/README.md
+++ b/app/server/README.md
@@ -480,6 +480,7 @@ Because the body carries audio rather than JSON, parameters are query parameters
 | `channels` | `1` | interleaved channel count |
 | `sample_format` | `s16le` | `s16le` or `f32le` |
 | `language` | unset | passed through to the model |
+| `prompt` | unset | URL-encoded recognition context (hotwords, spellings), same as the multipart `prompt` field |
 | `busy_timeout_ms` | model policy | how long to wait for the model lock, as elsewhere; clamped by the configured ceiling, so a request can shorten its own wait but never weaken the guard |
 
 ```bash
diff --git a/app/server/runtime.cpp b/app/server/runtime.cpp
index 4f7041b3..03a504f6 100644
--- a/app/server/runtime.cpp
+++ b/app/server/runtime.cpp
@@ -3018,10 +3018,15 @@ HttpResponse ServerState::handle_transcription_live(const HttpRequest & request)
         audio_contract.sample_rate = sample_rate;
         audio_contract.channels = channels;
         task_request.audio_input = std::move(audio_contract);
-        const std::string language = query_param(request.query, "language");
+        const std::string language = decoded_query_param(request.query, "language");
+        // Recognition-context biasing (hotwords), same meaning as the multipart
+        // route's `prompt` field; URL-encoded because it rides in the query.
+        const std::string prompt = decoded_query_param(request.query, "prompt");
         if (!language.empty()) {
             task_request.options["language"] = language;
-            task_request.text_input = engine::runtime::Transcript{std::string(), language};
+        }
+        if (!language.empty() || !prompt.empty()) {
+            task_request.text_input = engine::runtime::Transcript{prompt, language};
         }
         task_request = apply_default_request_options(model, std::move(task_request));
     } catch (const std::runtime_error & ex) {
diff --git a/docs/community_models/r2t2.md b/docs/community_models/r2t2.md
index bc1939f3..815b4a89 100644
--- a/docs/community_models/r2t2.md
+++ b/docs/community_models/r2t2.md
@@ -247,6 +247,8 @@ ffmpeg -f avfoundation -i ":0" -ar 16000 -ac 1 -f s16le - \
       'http://127.0.0.1:8488/v1/audio/transcriptions/live?model=r2t2-asr-stream&sample_rate=16000&channels=1&sample_format=s16le'
 ```
 
+Add `&prompt=<url-encoded hotwords>` to bias the live stream the same way `--text` does on the CLI.
+
 ## GGUF checkpoints
 
 GGUF is supported for Q8_0 and higher precision. `f16`, `q8_0`, and the native