fix(docs/retrievers): correct title and factory method calls

- retriever.md: rename title Retrievers → Retriever
- All files: replace incorrect mf.Retriever.bm25() / mf.Retriever.wikipedia()
  with the actual API — mf.Retriever.lexical("bm25") / mf.Retriever.web("wikipedia")

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: vilsonrodrigues

docs/learn/retrievers/bm25.md

@@ -21,7 +21,7 @@ All three share the same `add()` / `__call__()` / `acall()` interface.
     ```python
     import msgflux as mf
 
-    retriever = mf.Retriever.bm25()
+    retriever = mf.Retriever.lexical("bm25")
 
     retriever.add([
         "Python is a high-level programming language.",
@@ -49,7 +49,7 @@ Pure Python implementation. No external packages required.
 ```python
 import msgflux as mf
 
-retriever = mf.Retriever.bm25(k1=1.5, b=0.75)
+retriever = mf.Retriever.lexical("bm25", k1=1.5, b=0.75)
 ```
 
 ### `bm25s` — High-performance (scipy sparse matrices)
@@ -59,7 +59,7 @@ Uses the [`bm25s`](https://github.com/xhluca/bm25s) library. Significantly faste
 ```python
 import msgflux as mf
 
-retriever = mf.Retriever.bm25s(
+retriever = mf.Retriever.lexical("bm25s",
     k1=1.5,
     b=0.75,
     method="lucene",   # "lucene" | "robertson" | "atire" | "bm25l" | "bm25+"
@@ -76,7 +76,7 @@ Uses [`rank-bm25`](https://github.com/dorianbrown/rank_bm25), a popular BM25 lib
 ```python
 import msgflux as mf
 
-retriever = mf.Retriever.rank_bm25(k1=1.5, b=0.75)
+retriever = mf.Retriever.lexical("rank_bm25", k1=1.5, b=0.75)
 ```
 
 Install: `pip install rank-bm25`
@@ -101,7 +101,7 @@ Call `.add()` with a list of strings before searching. Documents are indexed inc
 ```python
 import msgflux as mf
 
-retriever = mf.Retriever.bm25()
+retriever = mf.Retriever.lexical("bm25")
 
 # First batch
 retriever.add([
@@ -136,7 +136,7 @@ response = retriever(
     ```python
     import msgflux as mf
 
-    retriever = mf.Retriever.bm25()
+    retriever = mf.Retriever.lexical("bm25")
     retriever.add([
         "Deep learning uses neural networks with many layers.",
         "The weather today is sunny and warm.",
@@ -160,7 +160,7 @@ response = retriever(
 ```python
 import msgflux as mf
 
-retriever = mf.Retriever.bm25()
+retriever = mf.Retriever.lexical("bm25")
 retriever.add([
     "Python is great for data science.",
     "Java is widely used in enterprise applications.",
@@ -184,7 +184,7 @@ Inspect how scores are distributed across your corpus for a given query:
 ```python
 import msgflux as mf
 
-retriever = mf.Retriever.bm25()
+retriever = mf.Retriever.lexical("bm25")
 retriever.add([
     "Machine learning automates pattern recognition.",
     "Deep learning is a subset of machine learning.",
@@ -211,7 +211,7 @@ Useful for choosing a `threshold` value — e.g. use `mean_score` to filter out
 ```python
 import msgflux as mf
 
-retriever = mf.Retriever.bm25()
+retriever = mf.Retriever.lexical("bm25")
 retriever.add([
     "Async Python enables non-blocking I/O.",
     "Asyncio is the standard async library.",

docs/learn/retrievers/retriever.md

@@ -1,4 +1,4 @@
-# Retrievers
+# Retriever
 
 ## ✦₊⁺ Overview
 
@@ -20,7 +20,7 @@ There are two retriever families:
     ```python
     import msgflux as mf
 
-    retriever = mf.Retriever.bm25()
+    retriever = mf.Retriever.lexical("bm25")
 
     retriever.add([
         "Python is a high-level programming language.",
@@ -40,7 +40,7 @@ There are two retriever families:
     ```python
     import msgflux as mf
 
-    retriever = mf.Retriever.wikipedia(summary=2)
+    retriever = mf.Retriever.web("wikipedia", summary=2)
 
     response = retriever("quantum entanglement")
 
@@ -91,7 +91,7 @@ Pass a list to search multiple queries in a single call. Results are returned in
 ```python
 import msgflux as mf
 
-retriever = mf.Retriever.bm25()
+retriever = mf.Retriever.lexical("bm25")
 retriever.add(["Doc A about Python", "Doc B about Java", "Doc C about Rust"])
 
 queries = ["Python language", "systems programming"]
@@ -112,7 +112,7 @@ All retrievers expose `.acall()` for async usage:
 ```python
 import msgflux as mf
 
-retriever = mf.Retriever.wikipedia(summary=3)
+retriever = mf.Retriever.web("wikipedia", summary=3)
 
 response = await retriever.acall("artificial intelligence", top_k=2)
 

docs/learn/retrievers/wikipedia.md

@@ -16,7 +16,7 @@ The `wikipedia` retriever fetches and returns Wikipedia article content at query
     ```python
     import msgflux as mf
 
-    retriever = mf.Retriever.wikipedia()
+    retriever = mf.Retriever.web("wikipedia")
 
     response = retriever("machine learning", top_k=2)
 
@@ -45,7 +45,7 @@ The `wikipedia` retriever fetches and returns Wikipedia article content at query
 ```python
 import msgflux as mf
 
-retriever = mf.Retriever.wikipedia(
+retriever = mf.Retriever.web("wikipedia",
     language="en",
     summary=3,           # Return only the first 3 sentences
     return_images=True,
@@ -64,7 +64,7 @@ By default, the full article content is returned. Set `summary` to an integer to
     ```python
     import msgflux as mf
 
-    retriever = mf.Retriever.wikipedia(summary=2)
+    retriever = mf.Retriever.web("wikipedia", summary=2)
 
     response = retriever("Eiffel Tower")
 
@@ -86,7 +86,7 @@ Enable `return_images=True` to get a list of image URLs from each article. Icons
     ```python
     import msgflux as mf
 
-    retriever = mf.Retriever.wikipedia(
+    retriever = mf.Retriever.web("wikipedia",
         return_images=True,
         max_return_images=3
     )
@@ -111,7 +111,7 @@ Set `language` to any Wikipedia language code:
         ```python
         import msgflux as mf
 
-        retriever = mf.Retriever.wikipedia(language="pt", summary=3)
+        retriever = mf.Retriever.web("wikipedia", language="pt", summary=3)
         response = retriever("inteligência artificial")
         print(response.data[0].results[0].data.content)
         ```
@@ -121,7 +121,7 @@ Set `language` to any Wikipedia language code:
         ```python
         import msgflux as mf
 
-        retriever = mf.Retriever.wikipedia(language="es", summary=3)
+        retriever = mf.Retriever.web("wikipedia", language="es", summary=3)
         response = retriever("aprendizaje automático")
         print(response.data[0].results[0].data.content)
         ```
@@ -131,7 +131,7 @@ Set `language` to any Wikipedia language code:
         ```python
         import msgflux as mf
 
-        retriever = mf.Retriever.wikipedia(language="fr", summary=3)
+        retriever = mf.Retriever.web("wikipedia", language="fr", summary=3)
         response = retriever("réseau de neurones")
         print(response.data[0].results[0].data.content)
         ```
@@ -143,7 +143,7 @@ Set `language` to any Wikipedia language code:
 ```python
 import msgflux as mf
 
-retriever = mf.Retriever.wikipedia(summary=2)
+retriever = mf.Retriever.web("wikipedia", summary=2)
 
 queries = ["Python programming", "Rust programming language", "Go programming"]
 response = retriever(queries, top_k=1)
@@ -165,7 +165,7 @@ A typical pattern: retrieve Wikipedia context, then pass it to an LLM:
     ```python
     import msgflux as mf
 
-    retriever = mf.Retriever.wikipedia(summary=5)
+    retriever = mf.Retriever.web("wikipedia", summary=5)
     chat = mf.Model.chat_completion("openai/gpt-4.1-mini")
 
     def answer_with_wikipedia(question: str) -> str:
@@ -191,7 +191,7 @@ A typical pattern: retrieve Wikipedia context, then pass it to an LLM:
 ```python
 import msgflux as mf
 
-retriever = mf.Retriever.wikipedia(summary=3)
+retriever = mf.Retriever.web("wikipedia", summary=3)
 
 queries = ["quantum computing", "photosynthesis", "black holes"]
 response = await retriever.acall(queries, top_k=1)