init: Engram v0.1 — native memory substrate for accumulating intelligence

Memory is not stored and retrieved — it is activated and propagated.
Implements the spreading activation model with salience decay, typed edges,
four memory tiers, and flat cosine vector search over a sled embedded store.

engram/bindings/kotlin/README.md

@@ -0,0 +1,21 @@
+# Kotlin / JVM Bindings (planned v0.2)
+
+Engram-core will be exposed to Kotlin/JVM via the C FFI layer in `engram-ffi`.
+
+## Approach
+
+Use JNA (Java Native Access) or JNI with the compiled `libengram_ffi` shared library:
+
+```
+cargo build --release -p engram-ffi
+# produces: target/release/libengram_ffi.dylib (macOS) / libengram_ffi.so (Linux)
+```
+
+The header file will be generated by `cbindgen` from `engram-ffi/src/lib.rs`.
+
+## Status
+
+Stub only. FFI functions exposed: `engram_open`, `engram_close`, `engram_node_count`,
+`engram_edge_count`, `engram_decay`, `engram_free_string`.
+
+Full Kotlin idiomatic wrapper (data classes, coroutine-friendly suspend functions) planned for v0.2.