@stackbone/cli 0.2.0 → 0.2.2
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/CHANGELOG.md +42 -0
- package/main.js +5418 -841
- package/package.json +2 -1
- package/stackbone-cli-0.2.2.tgz +0 -0
- package/stackbone-cli-0.2.0.tgz +0 -0
package/CHANGELOG.md
CHANGED
|
@@ -7,6 +7,48 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
|
|
|
7
7
|
|
|
8
8
|
## [Unreleased]
|
|
9
9
|
|
|
10
|
+
## [0.2.2] - 2026-07-06
|
|
11
|
+
|
|
12
|
+
### Fixed
|
|
13
|
+
|
|
14
|
+
- **The Studio "Logs" panel showed nothing in `stackbone dev`.** That panel
|
|
15
|
+
(live SSE tail) expected the agent to run as a separate child process, whose
|
|
16
|
+
stdout/stderr `startAgentProcess` captures and republishes — that's how
|
|
17
|
+
cloud and self-host work today. But locally, deep agents and workflow steps
|
|
18
|
+
run IN-PROCESS inside the CLI/emulator (no child to pipe), so a bare
|
|
19
|
+
`console.log` in agent or workflow code only ever reached the terminal
|
|
20
|
+
running `stackbone dev`, never the bus the panel reads from. The
|
|
21
|
+
orchestrator now bridges the process's own `console.log`/`info`/`debug`/
|
|
22
|
+
`warn`/`error` calls onto the same `agent.log` event the panel already
|
|
23
|
+
consumes (`bridgeProcessStdioToBus` in `@stackbone/agent-runner`), wired
|
|
24
|
+
right after the emulator boots. It patches `console.*` rather than the raw
|
|
25
|
+
`process.stdout`/`stderr` streams on purpose — `@clack/prompts` spinners
|
|
26
|
+
write their redraw frames straight to those streams, bypassing `console`,
|
|
27
|
+
so this leaves the boot/reload UI untouched.
|
|
28
|
+
|
|
29
|
+
## [0.2.1] - 2026-07-06
|
|
30
|
+
|
|
31
|
+
### Added
|
|
32
|
+
|
|
33
|
+
- **Scaffold support for `streamDeepAgent`.** `detectStreamingWorkflows`
|
|
34
|
+
recognizes a workflow that imports and calls `streamDeepAgent` from
|
|
35
|
+
`@stackbone/sdk/workflow`, and the `--calls` scaffold template now emits it —
|
|
36
|
+
new workflow-agent scaffolds stream token-by-token into the Studio
|
|
37
|
+
Playground out of the box.
|
|
38
|
+
|
|
39
|
+
### Fixed
|
|
40
|
+
|
|
41
|
+
- **`stackbone dev` boot output.** Deep-agent boot logs no longer print raw
|
|
42
|
+
`[deep-agents] …` lines to stdout, breaking the clack rail — they're routed
|
|
43
|
+
through the CLI's pino logger (only shown under `--verbose`) and
|
|
44
|
+
re-surfaced as one unified `Deep agents ready (…)` step. The emulator's own
|
|
45
|
+
boot step (an internal implementation detail) no longer gets its own
|
|
46
|
+
spinner line either.
|
|
47
|
+
- **`rag-ingest` default workflow no longer risks a workflow-VM crash.** The
|
|
48
|
+
staged-metadata read was a sibling helper at module scope that touched
|
|
49
|
+
`stackbone`, dragging the whole SDK client into the workflow bundle; it's
|
|
50
|
+
now inlined inside the `"use step"` function instead.
|
|
51
|
+
|
|
10
52
|
## [0.2.0] - 2026-07-02
|
|
11
53
|
|
|
12
54
|
### Added
|