wayang-ai 0.1.0 → 0.3.0
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.
- package/README.md +23 -4
- package/dist/cli.js +3497 -2429
- package/package.json +4 -3
package/README.md
CHANGED
|
@@ -64,11 +64,19 @@ Complex AI workflows need more than a single chat. Wayang acts as an intelligent
|
|
|
64
64
|
### Install
|
|
65
65
|
|
|
66
66
|
```bash
|
|
67
|
-
|
|
68
|
-
npm install
|
|
69
|
-
npm run build
|
|
67
|
+
npm install -g wayang-ai
|
|
70
68
|
```
|
|
71
69
|
|
|
70
|
+
After installation, the global command `wayang` (alias `waya`) is available in your terminal:
|
|
71
|
+
|
|
72
|
+
```bash
|
|
73
|
+
wayang # start a new session
|
|
74
|
+
waya # shorthand alias
|
|
75
|
+
wayang --resume # resume last session
|
|
76
|
+
```
|
|
77
|
+
|
|
78
|
+
On first run, an interactive setup wizard will guide you through configuring your LLM provider.
|
|
79
|
+
|
|
72
80
|
### Configure
|
|
73
81
|
|
|
74
82
|
On first run, Wayang launches an interactive setup wizard. Just fill in your LLM provider details.
|
|
@@ -80,7 +88,8 @@ Or create `~/.wayang.config.json` manually:
|
|
|
80
88
|
"providers": {
|
|
81
89
|
"my-model": {
|
|
82
90
|
"endpoint": "https://api.anthropic.com",
|
|
83
|
-
"modelName": "claude-sonnet-4-20250514"
|
|
91
|
+
"modelName": "claude-sonnet-4-20250514",
|
|
92
|
+
"apiKey": "xxx"
|
|
84
93
|
}
|
|
85
94
|
},
|
|
86
95
|
"controller": { "provider": "my-model" },
|
|
@@ -155,6 +164,16 @@ npm run test:watch # watch mode
|
|
|
155
164
|
npm run lint # type check + lint
|
|
156
165
|
```
|
|
157
166
|
|
|
167
|
+
## Roadmap
|
|
168
|
+
|
|
169
|
+
- [ ] **Controller ↔ Worker Chat** — Allow the controller to chat with a running worker and ask for human approval when needed
|
|
170
|
+
- [ ] **Persistent Workers Across Tasks** — Allow workers to survive task completion and receive follow-up tasks, retaining conversation context for iterative refinement without starting from scratch
|
|
171
|
+
- [ ] **Sessionless Long-Running Agent** — Rethink session design: always inherit context from the previous run + long-term memory, making Wayang a continuously running agent rather than a session-based tool
|
|
172
|
+
- [ ] **More Third-Party Worker Agents** — Expand integrations beyond Claude Code (e.g. Codex, Aider, Cursor agent)
|
|
173
|
+
- [ ] **TUI Improvements** — Better worker detail pages, richer status display, and overall UX polish
|
|
174
|
+
- [ ] **GUI** — Bring Wayang to a graphical interface beyond the terminal
|
|
175
|
+
- [ ] **Pluggable Worker Ecosystem** — Standardize a Worker plugin protocol (capabilities, tools, resources), build a plugin registry, and provide scaffolding so the community can develop and share custom Workers
|
|
176
|
+
|
|
158
177
|
## License
|
|
159
178
|
|
|
160
179
|
MIT
|