linear-mcp-fast 0.2.1__tar.gz → 0.2.3__tar.gz

{linear_mcp_fast-0.2.1 → linear_mcp_fast-0.2.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: linear-mcp-fast
-Version: 0.2.1
+Version: 0.2.3
 Summary: Fast MCP server for Linear - reads from Linear.app's local cache on macOS
 Author: everything-chalna
 License-Expression: MIT
@@ -26,12 +26,26 @@ Requires-Dist: brotli>=1.0.0
 
 Fast, read-only MCP server for Linear that reads from Linear.app's local cache on macOS.
 
-**Why?**
-- **Instant**: No API calls, reads directly from local IndexedDB cache
-- **Offline**: Works without internet
-- **Lower context**: Smaller responses for AI assistants
+## Why I Built This
 
-**Use with**: Official Linear MCP for write operations (comments, updates).
+While using the official Linear MCP with Claude Code, I noticed that **read operations consumed too much context**. Every issue query returned verbose responses with metadata I didn't need, eating into the AI's context window.
+
+The problem:
+- Official Linear MCP makes API calls for every read
+- Responses include excessive metadata (full user objects, workflow states, etc.)
+- Context window fills up quickly when exploring issues
+- Slower response times due to network latency
+
+My solution: **Read directly from Linear.app's local cache.**
+
+Linear.app (Electron) syncs all your data to a local IndexedDB. This MCP server reads from that cache, giving you:
+
+- **Zero API calls** - Instant reads from disk
+- **Smaller responses** - Only the fields you need
+- **Offline access** - Works without internet
+- **Faster iteration** - No rate limits, no latency
+
+**Use with**: Official Linear MCP for write operations (comments, status updates, issue creation).
 
 ## Installation
 

{linear_mcp_fast-0.2.1 → linear_mcp_fast-0.2.3}/README.md

@@ -2,12 +2,26 @@
 
 Fast, read-only MCP server for Linear that reads from Linear.app's local cache on macOS.
 
-**Why?**
-- **Instant**: No API calls, reads directly from local IndexedDB cache
-- **Offline**: Works without internet
-- **Lower context**: Smaller responses for AI assistants
+## Why I Built This
 
-**Use with**: Official Linear MCP for write operations (comments, updates).
+While using the official Linear MCP with Claude Code, I noticed that **read operations consumed too much context**. Every issue query returned verbose responses with metadata I didn't need, eating into the AI's context window.
+
+The problem:
+- Official Linear MCP makes API calls for every read
+- Responses include excessive metadata (full user objects, workflow states, etc.)
+- Context window fills up quickly when exploring issues
+- Slower response times due to network latency
+
+My solution: **Read directly from Linear.app's local cache.**
+
+Linear.app (Electron) syncs all your data to a local IndexedDB. This MCP server reads from that cache, giving you:
+
+- **Zero API calls** - Instant reads from disk
+- **Smaller responses** - Only the fields you need
+- **Offline access** - Works without internet
+- **Faster iteration** - No rate limits, no latency
+
+**Use with**: Official Linear MCP for write operations (comments, status updates, issue creation).
 
 ## Installation
 

{linear_mcp_fast-0.2.1 → linear_mcp_fast-0.2.3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "linear-mcp-fast"
-version = "0.2.1"
+version = "0.2.3"
 description = "Fast MCP server for Linear - reads from Linear.app's local cache on macOS"
 readme = "README.md"
 license = "MIT"

{linear_mcp_fast-0.2.1 → linear_mcp_fast-0.2.3}/src/linear_mcp_fast/reader.py

@@ -75,10 +75,13 @@ class LinearLocalReader:
     def _find_linear_db(
         self, wrapper: ccl_chromium_indexeddb.WrappedIndexDB
     ) -> ccl_chromium_indexeddb.WrappedDatabase:
-        """Find the main Linear database."""
+        """Find the main Linear database with data."""
         for db_id in wrapper.database_ids:
             if "linear_" in db_id.name and db_id.name != "linear_databases":
-                return wrapper[db_id.name, db_id.origin]
+                db = wrapper[db_id.name, db_id.origin]
+                # Skip empty databases
+                if list(db.object_store_names):
+                    return db
         raise ValueError("Could not find Linear database in IndexedDB")
 
     def _to_str(self, val: Any) -> str:

{linear_mcp_fast-0.2.1 → linear_mcp_fast-0.2.3}/src/linear_mcp_fast.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: linear-mcp-fast
-Version: 0.2.1
+Version: 0.2.3
 Summary: Fast MCP server for Linear - reads from Linear.app's local cache on macOS
 Author: everything-chalna
 License-Expression: MIT
@@ -26,12 +26,26 @@ Requires-Dist: brotli>=1.0.0
 
 Fast, read-only MCP server for Linear that reads from Linear.app's local cache on macOS.
 
-**Why?**
-- **Instant**: No API calls, reads directly from local IndexedDB cache
-- **Offline**: Works without internet
-- **Lower context**: Smaller responses for AI assistants
+## Why I Built This
 
-**Use with**: Official Linear MCP for write operations (comments, updates).
+While using the official Linear MCP with Claude Code, I noticed that **read operations consumed too much context**. Every issue query returned verbose responses with metadata I didn't need, eating into the AI's context window.
+
+The problem:
+- Official Linear MCP makes API calls for every read
+- Responses include excessive metadata (full user objects, workflow states, etc.)
+- Context window fills up quickly when exploring issues
+- Slower response times due to network latency
+
+My solution: **Read directly from Linear.app's local cache.**
+
+Linear.app (Electron) syncs all your data to a local IndexedDB. This MCP server reads from that cache, giving you:
+
+- **Zero API calls** - Instant reads from disk
+- **Smaller responses** - Only the fields you need
+- **Offline access** - Works without internet
+- **Faster iteration** - No rate limits, no latency
+
+**Use with**: Official Linear MCP for write operations (comments, status updates, issue creation).
 
 ## Installation