@olhapi/maestro-darwin-x64 0.1.3 → 0.1.5-rc.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/README.md +8 -2
- package/lib/maestro +0 -0
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -6,6 +6,8 @@ Website: [maestro.olhapi.com](https://maestro.olhapi.com/)
|
|
|
6
6
|
Repository: [github.com/olhapi/maestro](https://github.com/olhapi/maestro)
|
|
7
7
|
Docs: [maestro.olhapi.com/docs](https://maestro.olhapi.com/docs)
|
|
8
8
|
|
|
9
|
+
This project is inspired by [openai/symphony](https://github.com/openai/symphony).
|
|
10
|
+
|
|
9
11
|
It combines a SQLite-backed tracker, an orchestrator that reads `WORKFLOW.md`, a private MCP daemon bridged by `maestro mcp`, and an HTTP server that serves the embedded dashboard plus JSON/WebSocket APIs.
|
|
10
12
|
|
|
11
13
|
Maestro stays local-first. External work is translated into Maestro projects and issues through the CLI, the embedded dashboard, or MCP prompts, then supervised through the same local queue, runtime state, MCP tools, and dashboard surfaces.
|
|
@@ -289,13 +291,17 @@ Fresh `maestro workflow init --defaults` output currently defaults to:
|
|
|
289
291
|
- `agent.mode: app_server`
|
|
290
292
|
- `agent.dispatch_mode: parallel`
|
|
291
293
|
- `codex.command: codex app-server`
|
|
292
|
-
- `codex.expected_version: 0.
|
|
294
|
+
- `codex.expected_version: 0.116.0`
|
|
293
295
|
- `codex.approval_policy: never`
|
|
294
296
|
- `codex.initial_collaboration_mode: default` for fresh `app_server` threads
|
|
297
|
+
- `phases.review.enabled: true`
|
|
298
|
+
- `phases.done.enabled: true`
|
|
295
299
|
- runtime permission profiles now live in the DB per project/issue instead of `WORKFLOW.md`
|
|
296
300
|
|
|
297
301
|
`initial_collaboration_mode: default` keeps unattended runs execution-first for a fresh `app_server` thread. Use `plan` only when you explicitly want a plan-gated startup mode. Interactive approvals and `requestUserInput` prompts still depend on using a non-`never` approval policy, and those prompts are queued through the dashboard's global interrupt panel. Resumed threads and `stdio` runs do not use that startup-mode path.
|
|
298
302
|
|
|
303
|
+
Interactive `maestro workflow init` now walks through `workspace.root`, `codex.command`, `agent.mode`, `agent.dispatch_mode`, `agent.max_concurrent_agents`, `agent.max_turns`, and `agent.max_automatic_retries`, then asks for `codex.approval_policy` and `codex.initial_collaboration_mode` only for `app_server`. Those extra tuning knobs remain interactive-only; `--defaults` stays the stable scripted path, and the existing flags still override only the workspace root, Codex command, and agent mode.
|
|
304
|
+
|
|
299
305
|
Supported prompt-template variables are:
|
|
300
306
|
|
|
301
307
|
- `{{ issue.identifier }}`
|
|
@@ -309,7 +315,7 @@ Supported prompt-template variables are:
|
|
|
309
315
|
- `{{ attempt }}`
|
|
310
316
|
|
|
311
317
|
When a project has a description, Maestro's default implementation, review, and done prompts include it automatically under a `Project context:` section. Custom workflows can place `{{ project.description }}` wherever they want.
|
|
312
|
-
The default done prompt now focuses on merge-back, PR readiness, and blocker reporting instead of asking for a preview artifact.
|
|
318
|
+
The default done prompt now focuses on merge-back, worktree cleanup, PR readiness, and blocker reporting instead of asking for a preview artifact.
|
|
313
319
|
|
|
314
320
|
The checked-in [`WORKFLOW.md`](WORKFLOW.md) is this repository's own workflow example. It is not guaranteed to match fresh `workflow init` defaults exactly.
|
|
315
321
|
|
package/lib/maestro
CHANGED
|
Binary file
|