tagteam 0.3.0 → 0.4.2
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 +13 -3
- package/dist/index.js +452 -86
- package/dist/index.js.map +1 -1
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -81,12 +81,21 @@ This works for any agent, not just your active pair. Addressed agents receive th
|
|
|
81
81
|
|
|
82
82
|
### Discussion mode
|
|
83
83
|
|
|
84
|
-
Have your agents discuss a topic in rounds
|
|
84
|
+
Have your agents discuss a topic in structured rounds with research-backed prompting:
|
|
85
85
|
|
|
86
86
|
```bash
|
|
87
87
|
tagteam discuss "what's the best approach for caching in this app"
|
|
88
88
|
```
|
|
89
89
|
|
|
90
|
+
Discussions use a structured debate protocol informed by multi-agent debate research (Du et al. ICML 2024, ReConcile ACL 2024, CONSENSAGENT ACL Findings 2025, MAD EMNLP 2024):
|
|
91
|
+
|
|
92
|
+
- **Role differentiation** — each agent has a distinct role (Builder, Verifier, or Strategist) with specific focus areas, and sees an anonymized description of their peer's role
|
|
93
|
+
- **Anti-sycophancy rules** — agents must justify position changes with specific arguments and introduce novel content each round
|
|
94
|
+
- **Structured arguments** — debate responses follow Toulmin structure (claim, evidence, reasoning, caveats) with confidence and position-change markers
|
|
95
|
+
- **Steelman injection** — in round 2, agents are prompted to steelman the opposing view before responding
|
|
96
|
+
- **Smart termination** — discussions end early on mutual consensus, stale debates (no new arguments for 2 rounds), or cyclic position-swapping, with a clear status message explaining why
|
|
97
|
+
- **Context summarization** — rounds 3+ receive a compressed summary of earlier rounds plus full text of the latest exchange, keeping prompts focused
|
|
98
|
+
|
|
90
99
|
### Session management
|
|
91
100
|
|
|
92
101
|
```bash
|
|
@@ -128,7 +137,7 @@ tagteam config set agents claude,gemini
|
|
|
128
137
|
tagteam config set claude.model sonnet
|
|
129
138
|
tagteam config set codex.model gpt-5.3-codex
|
|
130
139
|
tagteam config set gemini.model gemini-2.5-pro
|
|
131
|
-
tagteam config set discussion.max_rounds
|
|
140
|
+
tagteam config set discussion.max_rounds 5
|
|
132
141
|
```
|
|
133
142
|
|
|
134
143
|
Default values:
|
|
@@ -139,7 +148,7 @@ Default values:
|
|
|
139
148
|
| `claude.model` | `sonnet` |
|
|
140
149
|
| `codex.model` | `gpt-5.3-codex` |
|
|
141
150
|
| `gemini.model` | `gemini-2.5-pro` |
|
|
142
|
-
| `discussion.max_rounds` | `
|
|
151
|
+
| `discussion.max_rounds` | `5` |
|
|
143
152
|
|
|
144
153
|
### CLI options
|
|
145
154
|
|
|
@@ -158,6 +167,7 @@ While in a session, type these slash commands:
|
|
|
158
167
|
- `/new` - Start a fresh session
|
|
159
168
|
- `/config` - Open the config editor (changes take effect immediately)
|
|
160
169
|
- `/copy` - Copy conversation to clipboard
|
|
170
|
+
- `/gist` - Create a private GitHub Gist of the conversation (requires [`gh`](https://cli.github.com))
|
|
161
171
|
- `/exit` - Exit the app
|
|
162
172
|
- `Escape` - Interrupt running agents
|
|
163
173
|
|