Correct paths

Author: johnsamuelwrites

docs/demos.md

@@ -11,33 +11,33 @@ The most compelling demo: the same agent logic written in English, French, and
 Japanese, running identically.  Proves that Multilingual is the only AI
 programming platform where agent code is idiomatic in any human language.
 
-- [`examples/agent_en.ml`](../examples/agent_en.ml) — English
-- [`examples/agent_fr.ml`](../examples/agent_fr.ml) — French
-- [`examples/agent_ja.ml`](../examples/agent_ja.ml) — Japanese
+- [`examples/agent_en.ml`](../../examples/agent_en.ml) — English
+- [`examples/agent_fr.ml`](../../examples/agent_fr.ml) — French
+- [`examples/agent_ja.ml`](../../examples/agent_ja.ml) — Japanese
 
 ### Reactive counter
 
 `observe var` + `on .change` + `canvas` — the simplest reactive web app.
 
-- [`examples/reactive_counter.ml`](../examples/reactive_counter.ml)
+- [`examples/reactive_counter.ml`](../../examples/reactive_counter.ml)
 
 ### Streaming chat (French)
 
 A streaming AI response bound to a reactive view, written in French.
 
-- [`examples/streaming_chat_fr.ml`](../examples/streaming_chat_fr.ml)
+- [`examples/streaming_chat_fr.ml`](../../examples/streaming_chat_fr.ml)
 
 ### Semantic search (Japanese)
 
 `embed` + `nearest` + `~=` semantic match across Japanese user input.
 
-- [`examples/semantic_search_ja.ml`](../examples/semantic_search_ja.ml)
+- [`examples/semantic_search_ja.ml`](../../examples/semantic_search_ja.ml)
 
 ### Multilingual AI dashboard
 
 `@agent`, reactive state, streaming output, and canvas composition.
 
-- [`examples/multilingual_dashboard.ml`](../examples/multilingual_dashboard.ml)
+- [`examples/multilingual_dashboard.ml`](../../examples/multilingual_dashboard.ml)
 
 ## CLI tools for Core 1.0
 

docs/playground.md

@@ -38,7 +38,7 @@ download required.
 multilingual build-wasm-bundle my_app.ml --out-dir dist/wasm
 ```
 
-Live demos using this approach are in [`examples/browser`](../examples/browser/).
+Live demos using this approach are in [`examples/browser`](../../examples/browser/).
 
 ## Core 1.0 example: the same agent in three languages