@zhivex-ai/agents 1.0.1 → 1.1.0

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.
Files changed (2) hide show
  1. package/README.md +5 -4
  2. package/package.json +2 -2
package/README.md CHANGED
@@ -40,17 +40,17 @@ console.log(result.state);
40
40
 
41
41
  ## Entry Points
42
42
 
43
- The package uses explicit entry points so production applications do not accidentally couple themselves to beta or experimental APIs:
43
+ The package uses explicit entry points so production applications can keep operational, realtime, beta, and testing dependencies intentional:
44
44
 
45
45
  | Import | Stability | Purpose |
46
46
  | --- | --- | --- |
47
47
  | `@zhivex-ai/agents` | Stable | Agent execution, tools, HITL, safety, streaming, handoffs, and subagents |
48
48
  | `@zhivex-ai/agents/ops` | Stable | Stores, memory, tracing, evaluation, replay, costs, and provider-support reports |
49
49
  | `@zhivex-ai/agents/beta` | Beta | Control plane, capsules, approval queues, ledgers, governance, and capability routing |
50
- | `@zhivex-ai/agents/realtime` | Experimental | Live/realtime agent streaming |
50
+ | `@zhivex-ai/agents/realtime` | Stable | Live/realtime agent streaming |
51
51
  | `@zhivex-ai/agents/testing` | Stable | Deterministic model and tool test doubles |
52
52
 
53
- Beta APIs may change between minor releases. Experimental APIs may change more frequently and should be isolated behind an application-owned adapter.
53
+ Beta APIs may change between minor releases. Provider-specific realtime options and upstream preview model availability are not promoted by the stable shared realtime contract.
54
54
 
55
55
  ## What This Package Covers
56
56
 
@@ -61,6 +61,7 @@ Beta APIs may change between minor releases. Experimental APIs may change more f
61
61
  - Durable harnesses: capsule fingerprints, app-provided execution-environment enforcement, and replay-visible context compaction.
62
62
  - Memory and stores from `/ops`: in-memory, file, SQLite, and Postgres run stores and memory stores.
63
63
  - Multi-agent patterns: handoffs, subagents as tools, parallel agent groups, and hierarchical traces.
64
+ - Durable subagents: stores with atomic idempotency claims reuse a completed child after a failed parent checkpoint instead of repeating child tools.
64
65
  - Production safety: stable safety policies and budget guards in the root; beta governance policies and audit records under `/beta`.
65
66
  - Observability and evaluation from `/ops`: trace collectors, run snapshots, replay, cost estimates, and evaluation fixtures.
66
67
  - Provider inspection from `/ops`, with beta capability routing and model selection under `/beta`.
@@ -201,7 +202,7 @@ Tier A means native agent building blocks such as approval-capable remote MCP or
201
202
 
202
203
  ## Realtime And Testing
203
204
 
204
- Keep experimental realtime use explicit:
205
+ Keep realtime dependencies explicit through the dedicated stable entry point:
205
206
 
206
207
  ```ts
207
208
  import { streamLiveAgent } from "@zhivex-ai/agents/realtime";
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@zhivex-ai/agents",
3
- "version": "1.0.1",
3
+ "version": "1.1.0",
4
4
  "description": "Agent-first facade for the Zhivex AI SDK runtime.",
5
5
  "license": "MIT",
6
6
  "type": "module",
@@ -57,6 +57,6 @@
57
57
  "access": "public"
58
58
  },
59
59
  "dependencies": {
60
- "@zhivex-ai/core": "^1.0.2"
60
+ "@zhivex-ai/core": "^1.4.0"
61
61
  }
62
62
  }