@reinsjs/tools-mcp 0.1.0 → 0.1.1

package/CHANGELOG.md

@@ -1,5 +1,13 @@
 # @reinsjs/tools-mcp
 
+## 0.1.1
+
+### Patch Changes
+
+- Docs only: the umbrella package is `@reinsjs/agent` (npm refused the bare name `reins` as too similar to `redis`). READMEs and the 0.1.0 changelog text now say so; no code changes.
+- Updated dependencies
+  - @reinsjs/core@0.1.1
+
 ## 0.1.0
 
 ### Minor Changes
@@ -8,11 +16,11 @@
 
   - **Timeline first.** Append-only event log with `schemaVersion` on every event and fail-closed upcast on read; projection decides what the model sees each turn; replay and fork from the log alone.
   - **A loop you can copy.** `runLoop` as an exported async generator: pause/resume is a return value with a small signed state, bounded retry on transient failures only while nothing has been written, tool pipeline `beforeTool → validate → approval → execute → afterTool`, sub-agent pause bubbling (`Interruption { kind: "subagent" }`).
-  - **Brain modules** (`@reinsjs/brain`): perception, compact (with folded-result manifest and `recall`; recommended on after measuring 100% recall at lower token cost on two model families), pins, spill, handoff, memory (`memory_20250818` shape, namespace isolation), approval (deny → ask → allow, fail-closed), budget (five limits, sub-agent usage included).
+  - **Brain modules** (`@reinsjs/brain`): perception, compact (with folded-result manifest and `recall`; recommended on after measuring 100% recall at lower token cost on two model families), pins, spill, handoff, memory (`memory_20250818` shape, namespace isolation), approval (deny → ask → allow, fail-closed), budget (five limits, sub-agent usage included), skills (Agent Skills with progressive disclosure: menu in the system prompt, `skill_read` to open one).
   - **Lowering** (`@reinsjs/lowering-pi`): Anthropic Messages and OpenAI Responses via pi-ai with a declared loss matrix, trust markers on untrusted content, cache-breakpoint handling.
   - **Serving** (`@reinsjs/server`, `@reinsjs/ui-agui`): Web-standard handler with SSE replay from `lastSeq`, per-session run lock, `authorizeSession` (only `true` allows), AG-UI as the first-class UI protocol.
   - **Stores**: SQLite (`node:sqlite` / `bun:sqlite`) and Postgres (`pg` / PGlite), conformance-tested; configurable memory table for per-role isolation.
-  - **Tools**: MCP servers as one socket (`@reinsjs/tools-mcp`, tools bound per run, annotations as defaults not permissions); `asTool(agent)` in `reins`.
+  - **Tools**: MCP servers as one socket (`@reinsjs/tools-mcp`, tools bound per run, annotations as defaults not permissions); `asTool(agent)` in `@reinsjs/agent`.
   - **TanStack AI** middleware (`@reinsjs/adapter-tanstack-ai`): the brain on TanStack's loop with the log as the single source of truth; idempotent client-message import; runtime check for the approval interrupt.
   - **Eval harness** (`@reinsjs/eval`): fixtures from real recordings, recorded tools, arm-vs-arm runner, metrics, the four-rule gate.
 

package/README.md

@@ -7,7 +7,7 @@ them exactly like in-process tools. The server never learns that reins exists.
 ```ts
 import { httpTransport, mcpTools } from "@reinsjs/tools-mcp"
 import { approval, spill } from "@reinsjs/brain"
-import { createAgent } from "reins"
+import { createAgent } from "@reinsjs/agent"
 
 const github = mcpTools({
   transport: httpTransport({

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@reinsjs/tools-mcp",
-  "version": "0.1.0",
+  "version": "0.1.1",
   "description": "reins — MCP servers as a Socket: tools/list once per run, tools/call as Tool.execute, annotations → risk / approval defaults. Streamable HTTP on the Web-standard entry, stdio under ./node",
   "keywords": [
     "agent",
@@ -37,13 +37,13 @@
   ],
   "dependencies": {
     "@modelcontextprotocol/client": "2.0.0",
-    "@reinsjs/core": "0.1.0"
+    "@reinsjs/core": "0.1.1"
   },
   "devDependencies": {
     "@modelcontextprotocol/server": "2.0.0",
     "@types/node": "^22.0.0",
     "zod": "^4.2.0",
-    "@reinsjs/brain": "0.1.0"
+    "@reinsjs/brain": "0.1.1"
   },
   "engines": {
     "node": ">=22"