smcreader 0.1.1 → 0.1.3

package/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "smcreader",
-  "version": "0.1.1",
+  "version": "0.1.3",
   "description": "SMC Reader (Save My Context Reader): large tool outputs are spilled to disk and replaced by a short pointer; the model pulls only the lines it needs via reader_ask / reader_peek.",
   "author": { "name": "bulatko" },
   "homepage": "https://github.com/bulatko/SMCReader",

package/.mcp.json

@@ -2,6 +2,6 @@
   "smcreader": {
     "type": "stdio",
     "command": "npx",
-    "args": ["-y", "--omit=optional", "--package=smcreader@0.1.1", "smcreader-mcp"]
+    "args": ["-y", "--omit=optional", "--package=smcreader@0.1.3", "smcreader-mcp"]
   }
 }

package/README.md

@@ -201,7 +201,10 @@ READER_E2E_PACKAGE=registry sh e2e/plugin.sh     # the published npm version ins
 install gets) and points its `.mcp.json` at `npm pack` of the working tree.
 
 **Releasing.** A marketplace install takes the plugin from the default branch, and its `.mcp.json`
-pins `smcreader@<version>` — so publish that version to npm *before* pushing the version bump.
+pins `smcreader@<version>` — so the version must be on npm *before* the bump reaches that branch.
+Bump the version in `package.json`, `.claude-plugin/plugin.json` and `.mcp.json` (`npm test` checks
+they agree), commit, push **only the tag** `v<version>`; the `Release` workflow publishes to npm
+(trusted publishing, with provenance). Then push the branch.
 
 The e2e verdict comes from `--output-format stream-json`, i.e. from what the model actually
 received, not from what it said: (1) the Bash `tool_result` contains `[READER ctx_` and not
@@ -210,6 +213,20 @@ received, not from what it said: (1) the Bash `tool_result` contains `[READER ct
 
 </details>
 
+## What it runs and sends
+
+- **The hook** (`bin/smcreader-hook`, no dependencies, no network) reads the tool output that Claude
+  Code passes to it. If the output is large, the hook writes it to `READER_DIR`
+  (default `~/.cache/smcreader`) and returns the pointer in its place.
+- **The MCP server** comes from the npm registry: the plugin starts it with `npx`
+  (`smcreader@<version>` and its dependencies), and later starts run from npm's cache.
+- **`reader_ask`** sends the document and your question to Claude Haiku through your own Claude
+  Code (`claude`, your credentials), like a subagent would. Those Haiku sessions keep their
+  transcripts under `READER_WORK_DIR`. `reader_open`, `reader_peek`, `reader_list` and
+  `reader_close` run locally.
+- Documents expire 48 h after their last use. There is no telemetry, and nothing else leaves your
+  machine.
+
 ## Limitations
 
 - Needs a Claude Code version that applies `hookSpecificOutput.updatedToolOutput` from command

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "smcreader",
-  "version": "0.1.1",
+  "version": "0.1.3",
   "description": "SMC Reader (Save My Context Reader): pull-model reading of large tool outputs for Claude Code. A PostToolUse hook spills big outputs to disk and leaves a short pointer; an MCP server returns only the lines or answers the model asks for.",
   "license": "MIT",
   "author": "bulatko",