mercury-agent 0.5.1 → 0.5.2

package/README.md

@@ -18,7 +18,7 @@ Mercury is a personal AI assistant that lives where you chat. It connects to Wha
 ## Prerequisites
 
 - **[Node.js](https://nodejs.org/)** >= 18 — Required for `npm install -g mercury-agent`
-- **[Bun](https://bun.sh)** >= 1.0 — JavaScript runtime used by Mercury
+- **[Bun](https://bun.sh)** >= 1.2 — JavaScript runtime used by Mercury
 - **[Docker](https://docs.docker.com/get-docker/)** — Required for running agent containers. On Linux, also install `docker-buildx` (`apt-get install docker-buildx`) — Docker Desktop includes it, but Docker Engine on Linux does not.
 - **Windows users:** Mercury runs best under [WSL2](https://learn.microsoft.com/en-us/windows/wsl/install). Install WSL2 with `wsl --install`, then install Bun and Docker inside it.
 
@@ -66,6 +66,8 @@ mercury run
 mercury service install
 ```
 
+> `mercury run` runs in the foreground (good for a first smoke test). For anything long-running, prefer `mercury service install` — it runs in the background and auto-restarts.
+
 ### Set up spaces and conversations
 
 Mercury discovers conversations from incoming traffic. They start **unlinked** — you assign them to **spaces** (memory boundaries).
@@ -310,7 +312,7 @@ Optional project file **`mercury.yaml`** (or **`mercury.yml`**) supplies non-sec
 | Variable | Default | Description |
 |----------|---------|-------------|
 | `MERCURY_DATA_DIR` | `.mercury` | Data directory |
-| `MERCURY_MAX_CONCURRENCY` | `3` | Max concurrent runs |
+| `MERCURY_MAX_CONCURRENCY` | `2` | Max concurrent runs |
 | `MERCURY_PORT` | `8787` | API port |
 | `MERCURY_BOT_USERNAME` | `mercury` | Bot display name |
 | `MERCURY_LOG_LEVEL` | `info` | Log level |
@@ -335,7 +337,7 @@ Supported OAuth providers: Anthropic, GitHub Copilot, Google Gemini CLI, Antigra
 | Variable | Default | Description |
 |----------|---------|-------------|
 | `MERCURY_MODEL_PROVIDER` | `anthropic` | Provider |
-| `MERCURY_MODEL` | `claude-opus-4-6` | Model |
+| `MERCURY_MODEL` | `claude-opus-4-8` | Model |
 | `MERCURY_ANTHROPIC_API_KEY` | — | API key |
 | `MERCURY_ANTHROPIC_OAUTH_TOKEN` | — | OAuth token (alternative) |
 
@@ -357,7 +359,7 @@ Supported OAuth providers: Anthropic, GitHub Copilot, Google Gemini CLI, Antigra
 
 | Variable | Default | Description |
 |----------|---------|-------------|
-| `MERCURY_AGENT_CONTAINER_IMAGE` | `ghcr.io/avishai-tsabari/mercury-agent:latest` | Container image |
+| `MERCURY_AGENT_IMAGE` | `ghcr.io/avishai-tsabari/mercury-agent:latest` | Container image |
 | `MERCURY_CONTAINER_TIMEOUT_MS` | `300000` | Container timeout (5 min) |
 | `MERCURY_CONTAINER_RUNTIME` | `runc` | `runc` (default) or `runsc` ([gVisor](https://gvisor.dev)) |
 | `MERCURY_CONTAINER_BWRAP_DOCKER_COMPAT` | `false` | Set `true` on Linux Docker Engine (see note below) |
@@ -413,7 +415,6 @@ mrctl config set trigger_patterns "@Bot,Bot"
 
 ## Docs
 
-- **Remaining work:** [TODOS](docs/TODOS.md) — Security, reliability, and ops gaps
 - **Platform setup:** [WhatsApp](docs/setup-whatsapp.md) · [Discord](docs/setup-discord.md) · [Slack](docs/setup-slack.md)
 - [Configuration](docs/configuration.md) (mercury.yaml + env) · [PRD: config load](docs/prd-config-load.md)
 - [Authentication](docs/auth/overview.md)

package/docs/runbooks/publish-checklist.md

@@ -15,11 +15,19 @@ git add package.json
 git commit -m "chore: bump version to $(node -p 'require("./package.json").version')"
 git tag "v$(node -p 'require("./package.json").version')"
 git push && git push --tags
-npm publish
+
+# Publish to npm manually. npm 2FA requires an interactive one-time
+# password, so this cannot run in CI — pass it inline (it expires in ~30s):
+npm publish --otp=<code-from-authenticator>
 ```
 
 Always tag and publish together — `package.json` is the single source of truth for both npm and GitHub.
 
+**CI does not publish to npm.** Pushing a `v*` tag runs `bun run check` and creates a
+GitHub Release (`.github/workflows/release.yml`), but the npm upload is a manual step
+done by a maintainer from a local machine. Verify afterward with
+`npm view mercury-agent version`.
+
 ## Beta testing on another machine
 
 Use `--tag beta` to publish a test version without affecting the stable `latest` tag.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "mercury-agent",
-  "version": "0.5.1",
+  "version": "0.5.2",
   "description": "Personal AI assistant for chat platforms (WhatsApp, Slack, Discord, Telegram)",
   "license": "MIT",
   "author": "Avishai Tsabari",

package/resources/profiles/coding/mercury-profile.yaml

@@ -10,6 +10,6 @@ env: []
 
 defaults:
   model_provider: anthropic
-  model: claude-sonnet-4-20250514
+  model: claude-opus-4-8
   trigger_patterns: "@Mercury,Mercury"
   bot_username: mercury

package/resources/profiles/general/mercury-profile.yaml

@@ -10,6 +10,6 @@ env: []
 
 defaults:
   model_provider: anthropic
-  model: claude-sonnet-4-20250514
+  model: claude-opus-4-8
   trigger_patterns: "@Mercury,Mercury"
   bot_username: mercury

package/resources/profiles/research/mercury-profile.yaml

@@ -10,6 +10,6 @@ env: []
 
 defaults:
   model_provider: anthropic
-  model: claude-sonnet-4-20250514
+  model: claude-opus-4-8
   trigger_patterns: "@Mercury,Mercury"
   bot_username: mercury

package/resources/templates/mercury.example.yaml

@@ -9,10 +9,10 @@
 # model:
 #   chain:
 #     - provider: anthropic
-#       model: claude-sonnet-4-20250514
+#       model: claude-opus-4-8
 #   # Legacy single primary + optional fallback (when chain is omitted):
 #   # provider: anthropic
-#   # model: claude-sonnet-4-20250514
+#   # model: claude-opus-4-8
 #   # fallback_provider: openai
 #   # fallback: gpt-4o-mini
 #   # max_retries_per_leg: 2

package/src/agent/container-entry.ts

@@ -132,7 +132,7 @@ function parseModelLegsFromEnv(): ModelLeg[] {
   return [
     {
       provider: process.env.MODEL_PROVIDER || "anthropic",
-      model: process.env.MODEL || "claude-opus-4-6",
+      model: process.env.MODEL || "claude-opus-4-8",
     },
   ];
 }

package/src/agent/container-runner.ts

@@ -29,8 +29,7 @@ import { ContainerError } from "./container-error.js";
  * launching the inner container detached (`docker create` + `docker start`, no
  * attach) is the only pattern that works through the Bun `fetch()`-based body-proxy
  * the cloud agent lane goes through — the proxy cannot carry Docker's hijacked
- * attach connection, so an attached run hangs to its idleTimeout (see
- * docs/debug/major/2026-05-25-agent-lane-docker-run-wait-hang-no-chat-response.md).
+ * attach connection, so an attached run hangs to its idleTimeout.
  */
 const INNER_IO_DIR = "/run/mercury-io";
 

package/src/cli/mercury.ts

@@ -721,7 +721,7 @@ program
       anthropic: {
         envKey: "MERCURY_ANTHROPIC_API_KEY",
         label: "Anthropic API key",
-        defaultModel: "claude-sonnet-4-20250514",
+        defaultModel: "claude-opus-4-8",
       },
       openai: {
         envKey: "MERCURY_OPENAI_API_KEY",

package/src/config.ts

@@ -62,7 +62,7 @@ const schema = z.object({
 
   // ─── AI Model ───────────────────────────────────────────────────────
   modelProvider: z.string().default("anthropic"),
-  model: z.string().default("claude-opus-4-6"),
+  model: z.string().default("claude-opus-4-8"),
   modelFallbackProvider: z.string().optional(),
   modelFallback: z.string().optional(),
   /** JSON array of `{ provider, model }`. When set, overrides legacy primary+fallback pair. */

package/src/main.ts

@@ -90,7 +90,7 @@ async function main() {
   // ─── Normalize data dir ownership ───────────────────────────────────────
   // Host base image's default user has drifted historically; if the volume
   // was written by a previous uid, SQLite can't reopen state.db for writes.
-  // Idempotent on matching ownership. See docs/debug/major/2026-04-16-host-base-image-user-drift.md.
+  // Idempotent on matching ownership.
   if (process.platform === "linux" && process.getuid?.() === 0) {
     const dataDirPath = resolveProjectPath(config.dataDir);
     if (existsSync(dataDirPath)) {