Fix OpenCode session cache source paths

CHANGELOG.md

@@ -38,6 +38,10 @@
   model aliases to priced Kimi K2 entries.
 
 ### Fixed (CLI)
+- **OpenCode session cache keys now use real database paths.** Discovery now
+  emits one source per `opencode*.db` file and the parser iterates root sessions
+  inside that database, so shared session-cache fingerprinting stats the actual
+  SQLite file instead of a synthetic `<dbPath>:<sessionId>` identifier.
 - **OpenCode child sessions are attributed to their root session.** The
   OpenCode parser now walks the unarchived `session.parent_id` subtree so
   child and grandchild agent sessions contribute token and tool usage under

docs/providers/opencode.md

@@ -4,19 +4,19 @@ OpenCode (sst/opencode).
 
 - **Source:** `src/providers/opencode.ts`
 - **Loading:** lazy (`src/providers/index.ts:59-75`)
-- **Test:** `tests/providers/opencode.test.ts` (676 lines, the largest provider test)
+- **Test:** `tests/providers/opencode.test.ts` (868 lines)
 
 ## Where it reads from
 
-Default `~/.local/share/opencode/` or `$XDG_DATA_HOME/opencode/`. The discovery walk picks up `opencode*.db` files (`opencode.ts:71-88`).
+Default `~/.local/share/opencode/` or `$XDG_DATA_HOME/opencode/`. The discovery walk picks up `opencode*.db` files (`opencode.ts:99-108`).
 
 ## Storage format
 
 SQLite.
 
 ## Caching
 
-None.
+Uses the shared session cache keyed by the real SQLite database file path.
 
 ## Deduplication
 
@@ -25,11 +25,12 @@ Per `<sessionId>:<messageId>`.
 ## Quirks
 
 - **Schema validation is loud.** When a required table is missing, the parser logs an actionable warning telling the user which table is gone and what version of OpenCode it expects. This is the right behavior; do not silently swallow these.
-- Source paths are encoded as `<dbPath>:<sessionId>`.
-- Discovery only emits root sessions (`parent_id IS NULL`) to avoid double
-  counting. Parsing a root session walks the unarchived `session.parent_id`
-  subtree, so child and grandchild agent sessions contribute their message,
-  token, and tool usage back to the root session.
+- Discovery emits one source per `opencode*.db` file so the shared cache can
+  fingerprint the actual SQLite file. The parser then iterates unarchived root
+  sessions (`parent_id IS NULL`) inside the database.
+- Parsing a root session walks the unarchived `session.parent_id` subtree, so
+  child and grandchild agent sessions contribute their message, token, and tool
+  usage back to the root session without double counting.
 - Each message's `parts` are indexed; preserving the order matters for reasoning-token correctness.
 - Tokens are reported across `input`, `output`, `reasoning`, `cache.read`, and `cache.write`. Anthropic semantics.
 - External MCP tools are stored as `<server>_<tool>` names (for example
@@ -39,6 +40,6 @@ Per `<sessionId>:<messageId>`.
 
 ## When fixing a bug here
 
-1. The 558-line test suite catches a lot. Run `npm test -- tests/providers/opencode.test.ts` before and after any change.
+1. The provider test suite catches a lot. Run `npm test -- tests/providers/opencode.test.ts` before and after any change.
 2. If the bug is "missing table" warning, do not catch and silence it. Either upgrade the version expectation in the parser or document the breaking schema change.
 3. If the bug is "reasoning tokens off by one", check the parts index ordering.