@mushi-mushi/mcp 0.16.0 → 0.16.1

package/CONTRIBUTING.md

@@ -58,9 +58,9 @@ packages/
   react-native/  React Native SDK (MIT)
   cli/           CLI tool (MIT)
   mcp/           MCP server for coding agents (MIT)
-  server/        Supabase Edge Functions (BSL)
-  agents/        Agentic fix pipeline (BSL)
-  verify/        Fix verification (BSL)
+  server/        Supabase Edge Functions (AGPLv3)
+  agents/        Agentic fix pipeline (AGPLv3)
+  verify/        Fix verification (AGPLv3)
 apps/
   admin/         Admin dashboard (React + Tailwind)
   docs/          Documentation site (Nextra — apps/docs)
@@ -97,8 +97,10 @@ Guardrails (run locally before pushing):
 pnpm check:docs-stats          # root README stat badges
 pnpm check:sdk-version-matrix  # apps/docs/content/sdks/index.mdx versions
 pnpm check:internal-doc-links  # dead /foo links in MDX
+pnpm check:onboarding-drift    # phantom MCP env vars / stale tool names
 pnpm check:bundle-docs         # @mushi-mushi/web size-limit vs README
 pnpm gen:llms-txt              # regenerate apps/docs/public/llms.txt after nav changes
+pnpm gen:mcp-tools-doc         # regenerate sdks/mcp-tools.generated.mdx from catalog.ts
 ```
 
 README changes on npm packages only appear on [npmjs.com](https://www.npmjs.com) after the next **Changesets publish**.
@@ -115,13 +117,30 @@ Select the affected packages, the semver bump type, and write a summary. The cha
 
 ## Release flow
 
-Releases are fully automated. Maintainers don't run `npm publish` by hand.
+Releases are fully automated once the version PR is merged. Maintainers don't run
+`npm publish` by hand for routine bumps. See **[`docs/DEPLOYMENT.md`](docs/DEPLOYMENT.md)**
+for the full production runbook (edge functions, admin SPA, docs site, migrations).
 
 1. PRs land on `master` with one or more changeset files in `.changeset/`.
-2. `release.yml` runs on every push to `master`. It opens (or updates) a `chore: version packages` PR that bumps every affected `package.json`, rolls up the changelogs, and deletes the consumed changesets.
-3. Merging that "Version Packages" PR re-fires `release.yml`. The publish step authenticates to npm via **OpenID Connect (OIDC) Trusted Publishers** — no long-lived `NPM_TOKEN` is exchanged — and every tarball ships with a **Sigstore provenance attestation** uploaded to the public transparency log.
+   Orphaned changesets (only ignored packages) fail `pnpm check:changeset-orphans`.
+2. `release.yml` runs on every push to `master`. It opens (or updates) a
+   **`chore: version packages`** PR on `changeset-release/master` that bumps
+   affected `package.json` files, rolls up changelogs, and deletes consumed
+   changesets.
+3. **Wait for CI on the version PR.** If checks don't appear (bot-branch
+   suppression), push an empty commit to `changeset-release/master`. If
+   `check:sdk-version-matrix` fails, run `pnpm gen:sdk-version-matrix` on that
+   branch and commit `apps/docs/content/sdks/index.mdx`.
+4. Merge the version PR. The publish step uses **OIDC Trusted Publishers** +
+   Sigstore provenance (with `NPM_TOKEN` only for brand-new package bootstrap).
+5. **Manually dispatch Release** unless a run starts within ~2 minutes:
+   **Actions → Release → Run workflow → `master`**. Squash-merges attributed to
+   `github-actions[bot]` usually suppress the auto `push` trigger — manual
+   dispatch is the normal path, not a fallback of last resort.
 
-If GitHub's anti-loop protection suppresses the auto re-fire (the squash merge can be attributed to `github-actions[bot]`), trigger the workflow manually: **Actions → release → Run workflow → master**.
+```bash
+gh workflow run release.yml --ref master
+```
 
 ### Adding a brand-new publishable package
 
@@ -168,7 +187,7 @@ chore: bump dependencies
 ## License
 
 - SDK packages are MIT — your contributions will be MIT-licensed
-- Server/agents/verify are BSL 1.1 — contributions to those packages fall under BSL
+- Server/agents/verify are AGPLv3 — contributions to those packages fall under AGPLv3
 
 ## Questions?
 

package/README.md

@@ -1,13 +1,8 @@
 # @mushi-mushi/mcp
 
-> **Sentry sees what code throws. Mushi sees what users feel — and closes the loop with AI.**
+> **Your AI wrote it. Mushi tells you why it broke.**
 
-[Model Context Protocol](https://spec.modelcontextprotocol.io/) server that wires Mushi's **evolution loop** into your AI coding agent. The loop is already running in your Mushi project:
-
-```
-User feels a bug → Mushi captures it → AI triages → AI opens a PR
-→ QA verifies → Judge scores → Lesson library remembers → next agent is smarter
-```
+[Model Context Protocol](https://spec.modelcontextprotocol.io/) server that wires Mushi into Cursor, Claude Code, and any MCP client. Pull fix context, run triage, and dispatch fixes without a second LLM vendor key.
 
 Wire it into Cursor, Claude Code, or any MCP client in one command:
 
@@ -16,7 +11,7 @@ npx mushi-mushi setup --ide cursor
 # or: npx mushi-mushi setup --ide claude
 ```
 
-That command reads `~/.mushirc`, writes `.cursor/mcp.json` with the `mushi` server block, and prints "Done — restart Cursor and ask: `list mushi tools`". No copy-pasting environment variables.
+That command reads CLI config at **`~/.config/mushi/config.json`** (legacy `~/.mushirc` auto-migrates), writes `.cursor/mcp.json` with the `mushi-<slug>` server block, and prints "Done — restart Cursor and ask: `list mushi tools`". No copy-pasting environment variables.
 
 ### Codebase Understand tools (indexed repos)
 
@@ -146,7 +141,7 @@ The Mushi backend now exposes the same tool catalog over the **Streamable HTTP**
 
 ### Feature groups (`?features=` / `MUSHI_FEATURES`)
 
-Full catalog is **79 tools**. New installs default to a lean set via:
+Full catalog is **71 tools**. New installs default to a lean set via:
 
 - **Stdio:** `MUSHI_FEATURES=triage,fixes,inventory,setup,docs`
 - **HTTP:** `?features=triage,fixes,inventory,setup,docs` on the MCP URL

package/dist/catalog.js

@@ -5,7 +5,7 @@ import {
   RESOURCE_CATALOG,
   TDD_TOOL_CATALOG,
   TOOL_CATALOG
-} from "./chunk-PZEGN4XD.js";
+} from "./chunk-UYZEIJ5G.js";
 export {
   ALL_SCOPES,
   CODEBASE_TOOL_CATALOG,

package/dist/{chunk-PZEGN4XD.js → chunk-UYZEIJ5G.js}

@@ -44,7 +44,7 @@ var TOOL_CATALOG = [
   {
     name: "get_fix_context",
     title: "Fix context bundle",
-    description: "Bundle the full context an agent needs to fix a bug: report detail, reproduction steps, component, root cause, ontology tags. One call instead of several.",
+    description: "Bundle the full context an agent needs to fix a bug: a paste-ready fixPrompt (plain-English diagnosis + reproduction + suggested fix + relevant code + blast radius), plus report detail, reproduction steps, component, root cause, and ontology tags. One call instead of several. No second LLM key needed.",
     scope: "mcp:read",
     hints: { readOnly: true, idempotent: true, openWorld: true },
     useCase: "Give me everything I need to fix this in one payload."

package/dist/index.js

@@ -51,7 +51,7 @@ var TOOL_CATALOG = [
   {
     name: "get_fix_context",
     title: "Fix context bundle",
-    description: "Bundle the full context an agent needs to fix a bug: report detail, reproduction steps, component, root cause, ontology tags. One call instead of several.",
+    description: "Bundle the full context an agent needs to fix a bug: a paste-ready fixPrompt (plain-English diagnosis + reproduction + suggested fix + relevant code + blast radius), plus report detail, reproduction steps, component, root cause, and ontology tags. One call instead of several. No second LLM key needed.",
     scope: "mcp:read",
     hints: { readOnly: true, idempotent: true, openWorld: true },
     useCase: "Give me everything I need to fix this in one payload."
@@ -1131,6 +1131,7 @@ function createMushiServer(config) {
       inputSchema: { reportId: z.string().describe("The report UUID to fix") },
       outputSchema: {
         report: z.unknown(),
+        fixPrompt: z.unknown(),
         reproductionSteps: z.unknown(),
         component: z.unknown(),
         rootCause: z.unknown(),
@@ -1141,6 +1142,10 @@ function createMushiServer(config) {
       const report = await apiCall(`/v1/admin/reports/${args.reportId}`);
       return jsonResult({
         report,
+        // Paste-ready fix prompt composed server-side by composeFixPacket()
+        // (diagnosis + repro + suggested fix + relevant code + blast radius).
+        // Hand this straight to the editing agent — no second LLM key needed.
+        fixPrompt: report.fix_packet ?? null,
         reproductionSteps: report.reproduction_steps ?? [],
         component: report.component,
         rootCause: report.stage2_analysis?.rootCause,

package/dist/server.js

@@ -2,7 +2,7 @@ import {
   CODEBASE_TOOL_CATALOG,
   TDD_TOOL_CATALOG,
   TOOL_CATALOG
-} from "./chunk-PZEGN4XD.js";
+} from "./chunk-UYZEIJ5G.js";
 import {
   MUSHI_SERVER_METADATA
 } from "./chunk-EHB2BGWX.js";
@@ -409,6 +409,7 @@ function createMushiServer(config) {
       inputSchema: { reportId: z.string().describe("The report UUID to fix") },
       outputSchema: {
         report: z.unknown(),
+        fixPrompt: z.unknown(),
         reproductionSteps: z.unknown(),
         component: z.unknown(),
         rootCause: z.unknown(),
@@ -419,6 +420,10 @@ function createMushiServer(config) {
       const report = await apiCall(`/v1/admin/reports/${args.reportId}`);
       return jsonResult({
         report,
+        // Paste-ready fix prompt composed server-side by composeFixPacket()
+        // (diagnosis + repro + suggested fix + relevant code + blast radius).
+        // Hand this straight to the editing agent — no second LLM key needed.
+        fixPrompt: report.fix_packet ?? null,
         reproductionSteps: report.reproduction_steps ?? [],
         component: report.component,
         rootCause: report.stage2_analysis?.rootCause,

package/package.json

@@ -1,6 +1,7 @@
 {
   "name": "@mushi-mushi/mcp",
-  "version": "0.16.0",
+  "mcpName": "io.github.kensaurus/mushi-mushi",
+  "version": "0.16.1",
   "license": "MIT",
   "description": "MCP server for Mushi Mushi — triage user-felt bugs, pull evidence (screenshots, console logs, network requests), check blast radius, and dispatch fix PRs, all from your AI coding agent",
   "type": "module",
@@ -42,7 +43,7 @@
   "dependencies": {
     "@modelcontextprotocol/sdk": "^1.29.0",
     "zod": "^4.4.2",
-    "@mushi-mushi/core": "^1.15.0"
+    "@mushi-mushi/core": "^1.16.0"
   },
   "devDependencies": {
     "@types/node": "^22.19.17",