nextclaw 0.6.31 → 0.6.32

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "nextclaw",
3
- "version": "0.6.31",
3
+ "version": "0.6.32",
4
4
  "description": "Lightweight personal AI assistant with CLI, multi-provider routing, and channel integrations.",
5
5
  "private": false,
6
6
  "type": "module",
@@ -38,9 +38,9 @@
38
38
  "dependencies": {
39
39
  "chokidar": "^3.6.0",
40
40
  "commander": "^12.1.0",
41
- "@nextclaw/core": "^0.6.24",
42
- "@nextclaw/server": "^0.4.13",
43
- "@nextclaw/openclaw-compat": "^0.1.17"
41
+ "@nextclaw/core": "^0.6.25",
42
+ "@nextclaw/server": "^0.4.14",
43
+ "@nextclaw/openclaw-compat": "^0.1.18"
44
44
  },
45
45
  "devDependencies": {
46
46
  "@types/node": "^20.17.6",
@@ -155,6 +155,9 @@ You can now configure OpenClaw-style multi-agent runtime behavior directly in th
155
155
  - `session.dmScope`: DM isolation strategy (`main` / `per-peer` / `per-channel-peer` / `per-account-channel-peer`)
156
156
  - `session.agentToAgent.maxPingPongTurns`: cap cross-agent ping-pong loops (`0` means block auto ping-pong)
157
157
 
158
+ > ⚠️ **Strict enum guard (OpenClaw-aligned):** `session.dmScope` accepts **only** these 4 values: `main`, `per-peer`, `per-channel-peer`, `per-account-channel-peer`.
159
+ > Any other value (for example `per-account-channel-peer-agent`) is invalid and must not be written.
160
+
158
161
  See full architecture details in [Multi-Agent Architecture](guides/multi-agent-architecture.md).
159
162
 
160
163
  Example:
@@ -327,6 +330,11 @@ For internal AI operations (same as other built-in capabilities):
327
330
  - Yes, the runtime registers the `gateway` tool (`config.get` / `config.schema` / `config.apply` / `config.patch`).
328
331
  - The AI can use it to manage the same config surface when you explicitly ask.
329
332
  - As with all config mutations, it follows the explicit-request rule (no silent self-mutation).
333
+ - **Required safe flow for AI config writes:**
334
+ 1. run `config.get` to read current config + hash;
335
+ 2. run `config.schema` and copy enum values exactly (no invented suffixes/variants);
336
+ 3. run `config.patch` with minimal patch;
337
+ 4. run `config.get` again to verify persisted value.
330
338
 
331
339
  ---
332
340