@rvry/mcp 0.12.2 → 1.0.0

package/README.md

@@ -15,13 +15,13 @@ The wizard will:
 2. Auto-detect supported clients on your machine
 3. Configure them automatically
 
-Supported clients: **Claude Code**, **Claude Desktop**, **Cursor**, **Gemini CLI**, **Codex**, **Anti-Gravity**
+Supported clients: **Claude Code**, **Claude Desktop**
 
 Restart any running clients after setup, and RVRY is ready.
 
 ### Manual Configuration
 
-For any MCP client that supports **stdio transport** (runs a local process), add this to your MCP config:
+For Claude Desktop or any other MCP client that supports **stdio transport**, add this to your MCP config:
 
 ```json
 {
@@ -37,33 +37,18 @@ For any MCP client that supports **stdio transport** (runs a local process), add
 
 Get your token at [rvry.ai/dashboard](https://rvry.ai/dashboard) after signing up.
 
-### HTTP Transport (Web-Based Clients)
-
-For clients that connect via HTTP URL (ChatGPT, Perplexity, etc.):
-
-**MCP Server URL:** `https://engine.rvry.ai/mcp`
-
-Authentication is handled via OAuth — you'll be prompted to sign in with Google when you first connect. No token configuration needed.
-
-### Compatibility
-
-RVRY supports both **stdio** and **HTTP** transports:
-
-| Transport | Client | Setup |
-|-----------|--------|-------|
-| stdio | Claude Code, Claude Desktop, Cursor, Gemini CLI, Codex, Anti-Gravity | `npx --yes --package @rvry/mcp@latest rvry-mcp setup` |
-| stdio | Any client with `command` + `args` config | Manual JSON config (see above) |
-| HTTP | ChatGPT (custom GPTs), Perplexity, any HTTP MCP client | URL: `https://engine.rvry.ai/mcp` |
-
 ## Tools
 
 | Tool | What it does | Tier |
 |------|-------------|------|
-| **`RVRY_deepthink`** | Extended multi-round analysis. Catches assumptions, tests them, and doesn't let your AI wrap up until it's dealt with what it found. | All |
-| **`RVRY_think`** | Lighter single-session analysis for questions that need structure but not the full deepthink loop. | All |
-| **`RVRY_problem_solve`** | Structured decision-making. Forces orientation, anticipation, and evaluation before committing to a recommendation. | Pro+ |
-| **`RVRY_challenge`** | Adversarial stress-testing. Finds weaknesses, tests assumptions, and surfaces failure modes in a proposal before you commit. | Pro+ |
-| **`RVRY_meta`** | Sustained metacognitive observation. Examines how your AI is thinking, not just what it's thinking. | Max |
+| **`rvry`** | Lighter single-session analysis for questions that need structure but not the full deepthink loop. (Renamed from `think` in 1.0.0.) | All |
+| **`deepthink`** | Extended multi-round analysis. Catches assumptions, tests them, and doesn't let your AI wrap up until it's dealt with what it found. | All |
+| **`shimmer`** | Three-round reflective answering: a protected observation round surfaces commitments the answer must honor, then the answer is produced under them, then at most one resolution round. | All |
+| **`problem_solve`** | Structured decision-making. Forces orientation, anticipation, and evaluation before committing to a recommendation. | Pro+ |
+| **`challenge`** | Adversarial stress-testing. Finds weaknesses, tests assumptions, and surfaces failure modes in a proposal before you commit. | Pro+ |
+| **`meta`** | Sustained metacognitive observation. Examines how your AI is thinking, not just what it's thinking. | Max |
+
+A `think` tombstone tool remains for backward compatibility — calling it returns a notice pointing at `rvry`. It will be removed at 2.0.0.
 
 ### Parameters
 
@@ -71,35 +56,42 @@ All tools accept:
 
 | Parameter | Type | Required | Description |
 |-----------|------|----------|-------------|
-| `input` | string | Yes | Your question (new session) or analysis (continuation) |
-| `sessionId` | string | No | Session ID for continuing an existing session |
-| `requiresCurrentData` | boolean | No | Set to `true` when the question involves facts, policies, prices, or events that may have changed. Seeds a verification constraint that persists across all rounds. |
-| `skipShimmer` | boolean | No | Set to `true` to skip the self-observation priming phase and go directly to analysis. Default `false`. Not available on `RVRY_meta`. |
+| `input` | string | Yes | Your question (new session) or analysis findings (continuation). |
+| `sessionId` | string | No | Session ID for continuing an existing session. |
+| `skipScoping` | boolean | No | Skip the scoping questions phase and begin analysis immediately. |
+| `userConstraints` | string[] | No | User-stated directives or requirements to track across rounds. |
+| `requiresCurrentData` | boolean | No | Set to `true` when the question depends on facts, policies, prices, or events that may have changed. Seeds a verification constraint that persists across all rounds. |
+| `record` | boolean | No | When `true` on every call, the local client archives each round's body to `.rvry/<op>/<slug>/round-N-<mode>.md`. Passed from slash commands via the `--record` flag. |
+| `harvest` | object | No | Final-call-only. The client includes its synthesis (`title`, `summary`, `keyFindings`, `shimmer`, `openQuestions`, `followUps`, plus operation-specific fields) so it can be persisted to the local session file. |
+| `rounds` | array | No | Final-call-only, present only when `--record` was set. Array of `{round, mode, body}` for each analytical round. |
 
 **When to use `requiresCurrentData`:** Any question about regulations, pricing, legal rulings, market conditions, API specifications, or current events. The engine will require the model to explicitly classify every cited fact as either verified-current or training-data-dependent before concluding.
 
-**When to use `skipShimmer`:** If you want to minimize round count and have a well-scoped, specific question, skipping shimmer removes one round of priming. For exploratory or open-ended questions, leave shimmer enabled — the self-observation priming improves cold-start depth.
+**SHIMMER display:** SHIMMER is always round 1. The model displays `RVRY SHIMMER` with the revery definition before its self-observation — this is the priming phase, where the model observes its own processing before structured analysis begins.
 
-**SHIMMER display:** When shimmer runs, the model displays `RVRY SHIMMER` with the revery definition before its self-observation. This is the priming phase — the model observes its own processing before structured analysis begins.
+**Client-owned harvest (Path B):** On the call that completes the session, the client constructs and passes a `harvest` object describing the final synthesis; the local writer renders `.rvry/<op>/<slug>/harvest.md` from it. With `--record`, each round is also written as `round-N-<mode>.md` in that folder.
 
 ## Slash Commands (Claude Code)
 
-When you run `npx --yes --package @rvry/mcp@latest rvry-mcp setup` and select Claude Code, the wizard installs 4 slash commands:
+When you run `npx --yes --package @rvry/mcp@latest rvry-mcp setup` and select Claude Code, the wizard installs these slash commands:
 
 | Command | Tool | What it does |
 |---------|------|-------------|
-| `/deepthink` | RVRY_deepthink | Deep multi-round analysis |
-| `/problem-solve` | RVRY_problem_solve | Structured decision-making |
-| `/challenge` | RVRY_challenge | Adversarial stress-testing |
-| `/meta` | RVRY_meta | Metacognitive observation |
+| `/rvry` | rvry | Structured single-session analysis with a Difficulty Gate |
+| `/deepthink` | deepthink | Deep multi-round analysis |
+| `/shimmer` | shimmer | Reflective answering under surfaced commitments |
+| `/problem-solve` | problem_solve | Structured decision-making |
+| `/challenge` | challenge | Adversarial stress-testing |
+| `/meta` | meta | Metacognitive observation |
+| `/think` | rvry | Deprecated alias for `/rvry` (removed next major) |
 
 Usage: type `/deepthink [your question]` in Claude Code. Each command routes to the corresponding RVRY MCP tool and includes rendering instructions so analysis appears as visible markdown rather than hidden in tool calls.
 
-**Updating commands:** Existing slash command files are never overwritten. To get the latest specs, delete the old files in `~/.claude/commands/` and re-run setup.
+**`--record` flag:** Append `--record` to any command (e.g. `/deepthink --record [question]`) to archive per-round bodies under `.rvry/<op>/<slug>/round-N-<mode>.md` alongside the harvest file.
 
-`RVRY_think` is available as a tool, but the setup wizard does not install a `/think` slash command. Use it from clients that show the full MCP tool list or invoke it manually.
+**Updating commands:** Re-running setup refreshes the RVRY-owned command files. Any pre-existing file is backed up to `<name>.md.bak` before it is overwritten.
 
-Commands are installed to `~/.claude/commands/`. Existing files are never overwritten — re-running setup is safe.
+Commands are installed to `~/.claude/commands/`.
 
 ## How it Works
 
@@ -115,8 +107,8 @@ On hard questions, a smaller model with RVRY outperforms a bigger model without
 
 | Plan | Sessions | Tools | Price |
 |------|----------|-------|-------|
-| Free | 5/month | deepthink | $0 |
-| Pro | 21/month | deepthink, problem_solve, challenge | $20/mo |
+| Free | 5/month | deepthink, think | $0 |
+| Pro | 21/month | deepthink, think, problem_solve, challenge | $20/mo |
 | Max | Unlimited | All tools including meta | $99/mo |
 
 Sign up at [rvry.ai](https://rvry.ai).

package/dist/client.d.ts

@@ -2,7 +2,7 @@
  * RVRY MCP Client -- HTTP client for the RVRY engine /api/v1/think endpoint.
  */
 /** Valid tool names for the RVRY engine */
-export type RvryTool = 'deepthink' | 'problem_solve' | 'think' | 'challenge' | 'meta';
+export type RvryTool = 'deepthink' | 'problem_solve' | 'think' | 'challenge' | 'meta' | 'shimmer';
 export interface ScopingQuestion {
     question: string;
     options: Array<{
@@ -13,7 +13,7 @@ export interface ScopingQuestion {
 }
 export interface ThinkResponse {
     sessionId: string;
-    status: 'scoping' | 'orient' | 'active' | 'complete';
+    status: 'scoping' | 'shimmer' | 'orient' | 'active' | 'complete';
     round: number;
     prompt: string;
     instruction: string;

package/dist/index.d.ts

@@ -5,7 +5,9 @@
  * Thin client that proxies tool calls to the RVRY engine HTTP API.
  * Auth via RVRY_TOKEN env var (static rvry_ token).
  *
- * Exposes 5 tools (RVRY_deepthink, RVRY_problem_solve, RVRY_think, RVRY_challenge, RVRY_meta) + backward-compat aliases + 5 MCP Prompts.
+ * Exposes 6 tools (rvry, deepthink, problem_solve, challenge, meta, shimmer)
+ * + RVRY_* backward-compat aliases + a `think` tombstone (renamed to rvry,
+ * removed at 2.0.0) + MCP Prompts.
  *
  * Also supports `npx --yes --package @rvry/mcp@latest rvry-mcp setup` to install Claude Code commands.
  */
@@ -14,11 +16,20 @@ import { type RvryTool, type ThinkResponse } from './client.js';
 /** Cache original question per session for response enrichment */
 declare const questionCache: Map<string, string>;
 /**
- * Map tool name to RvryTool.
- * Primary entries use rvry_ prefix; unprefixed entries are backward-compat aliases.
+ * Map tool name to RvryTool (the engine wire name).
+ * `rvry` maps to the engine's think operation. `think`/`RVRY_think` are
+ * tombstones handled before this map is consulted.
  */
 declare const TOOL_MAP: Record<string, RvryTool>;
+/** Non-error tombstone notice for the renamed think tool. Removed at 2.0.0. */
+declare const THINK_TOMBSTONE_NOTICE = "The think tool has been renamed to rvry. Call the rvry tool with the same arguments (input, sessionId, ...).";
 declare const TOOL_DEFS: Tool[];
-/** Strip response fields based on status for context-efficient MCP responses */
-declare function stripResponse(result: ThinkResponse, question: string): Record<string, unknown>;
-export { TOOL_MAP, TOOL_DEFS, stripResponse, questionCache };
+/**
+ * Strip response fields based on status for context-efficient MCP responses.
+ *
+ * `toolName` is the public MCP tool name of the call in scope. The 'shimmer'
+ * status passthrough applies ONLY when the called tool is shimmer (R-6):
+ * legacy ops keep dropping prompt/instruction on their priming round.
+ */
+declare function stripResponse(result: ThinkResponse, question: string, toolName?: string): Record<string, unknown>;
+export { TOOL_MAP, TOOL_DEFS, stripResponse, questionCache, THINK_TOMBSTONE_NOTICE };