Skip to content

Commit 953bb8d

Browse files
that-github-userunknownclaude
authored
Polish README with badges, architecture diagram, and quick-links (#12)
- Add shields.io badges: CI status, license, version, PRs welcome - Add centered tagline and quick-links navigation bar - Add ASCII architecture diagram showing parallel agent flow - Restructure with Quick Start section up front - Add apply command to usage examples - Remove redundant License section (LICENSE file exists) Closes #6 Co-authored-by: unknown <that-github-user@github.com> Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
1 parent 8e94cfd commit 953bb8d

File tree

1 file changed

+71
-22
lines changed

1 file changed

+71
-22
lines changed

README.md

Lines changed: 71 additions & 22 deletions
Original file line numberDiff line numberDiff line change
@@ -2,24 +2,79 @@
22
<img src="assets/logo.png" alt="thinktank" width="200" />
33
</p>
44

5-
# thinktank
5+
<h3 align="center">Ensemble AI coding — multiple agents, one best answer</h3>
66

7-
Ensemble AI coding. Run N parallel Claude Code agents on the same task, then select the best result via test execution and convergence analysis.
7+
<p align="center">
8+
<a href="https://github.com/that-github-user/thinktank/actions/workflows/ci.yml"><img src="https://img.shields.io/github/actions/workflow/status/that-github-user/thinktank/ci.yml?branch=main&style=for-the-badge&label=CI" alt="CI"></a>
9+
<a href="https://github.com/that-github-user/thinktank/blob/main/LICENSE"><img src="https://img.shields.io/github/license/that-github-user/thinktank?style=for-the-badge" alt="License"></a>
10+
<a href="https://github.com/that-github-user/thinktank/releases"><img src="https://img.shields.io/github/v/release/that-github-user/thinktank?style=for-the-badge&include_prereleases&label=version" alt="Version"></a>
11+
<a href="https://github.com/that-github-user/thinktank/pulls"><img src="https://img.shields.io/badge/PRs-welcome-brightgreen?style=for-the-badge" alt="PRs Welcome"></a>
12+
</p>
13+
14+
<p align="center">
15+
<a href="#quick-start">Quick Start</a> &middot;
16+
<a href="#how-it-works">How It Works</a> &middot;
17+
<a href="CONTRIBUTING.md">Contributing</a> &middot;
18+
<a href="#references">References</a>
19+
</p>
20+
21+
---
22+
23+
Run N parallel Claude Code agents on the same task, then select the best result via test execution and convergence analysis. Based on the principle that **the aggregate of independent attempts outperforms any single attempt** — proven in [ensemble ML](https://en.wikipedia.org/wiki/Ensemble_learning), [superforecasting](https://en.wikipedia.org/wiki/Superforecasting), and [LLM code generation research](#references).
24+
25+
## Quick start
26+
27+
```bash
28+
npm install -g thinktank
29+
30+
# Run 3 parallel agents on a task
31+
thinktank run "fix the authentication bypass"
32+
33+
# Run 5 agents with test verification
34+
thinktank run "fix the race condition" -n 5 -t "npm test"
35+
36+
# Apply the best result
37+
thinktank apply
38+
```
839

9-
Based on the principle that **the aggregate of independent attempts outperforms any single attempt** — proven in [ensemble ML](https://en.wikipedia.org/wiki/Ensemble_learning), [superforecasting](https://en.wikipedia.org/wiki/Superforecasting), and [LLM code generation research](#references).
40+
Requires [Claude Code CLI](https://docs.anthropic.com/en/docs/claude-code) installed and authenticated.
1041

1142
## How it works
1243

1344
```
14-
thinktank run "fix the authentication bypass" -n 5 -t "npm test"
45+
┌─────────────┐
46+
│ Your task │
47+
└──────┬──────┘
48+
49+
┌────────────┼────────────┐
50+
│ │ │
51+
▼ ▼ ▼
52+
┌──────────┐ ┌──────────┐ ┌──────────┐
53+
│ Agent #1 │ │ Agent #2 │ │ Agent #3 │
54+
│ worktree │ │ worktree │ │ worktree │
55+
└────┬─────┘ └────┬─────┘ └────┬─────┘
56+
│ │ │
57+
▼ ▼ ▼
58+
┌──────────────────────────────────────┐
59+
│ Test & Convergence │
60+
│ ┌─────────┐ ┌──────────────────┐ │
61+
│ │ npm test │ │ Agents 1,3 agree │ │
62+
│ └─────────┘ └──────────────────┘ │
63+
└───────────────────┬──────────────────┘
64+
65+
66+
┌─────────────────┐
67+
│ Best result │
68+
│ recommended │
69+
└─────────────────┘
1570
```
1671

1772
1. Spawns **N parallel Claude Code agents**, each in an isolated git worktree
1873
2. Each agent independently solves the task (no shared context = true independence)
1974
3. Runs your **test suite** on each result
2075
4. Analyzes **convergence** — did the agents agree on an approach?
2176
5. **Recommends** the best candidate (tests passing + consensus + smallest diff)
22-
6. You review and apply
77+
6. You review and `thinktank apply`
2378

2479
## Why this works
2580

@@ -31,25 +86,19 @@ Every model ever benchmarked shows **pass@5 >> pass@1**. The gap between "one at
3186
| Confidence | "Did it get it right?" | "4/5 agents agree — high confidence" |
3287
| Coverage | One approach explored | Multiple approaches, pick the best |
3388

89+
The key insight: **parallel attempts cost more tokens but not more time.** All agents run simultaneously.
90+
3491
## When to use it
3592

3693
- **High-stakes changes** — auth, payments, security, data migrations
3794
- **Ambiguous tasks** — multiple valid approaches, need to see the spread
3895
- **Complex refactors** — many files, easy to miss something
3996
- **Unfamiliar codebases** — agents might go the wrong direction
4097

41-
## Install
42-
43-
```bash
44-
npm install -g thinktank
45-
```
46-
47-
Requires [Claude Code CLI](https://docs.anthropic.com/en/docs/claude-code) installed and authenticated.
48-
4998
## Usage
5099

51100
```bash
52-
# Run 3 parallel agents (default)
101+
# Run with defaults (3 agents, sonnet model)
53102
thinktank run "add rate limiting to the API"
54103

55104
# Run 5 agents with test verification
@@ -58,11 +107,17 @@ thinktank run "fix the race condition in the cache layer" -n 5 -t "npm test"
58107
# Use a specific model
59108
thinktank run "migrate callbacks to async/await" --model opus -n 3
60109

110+
# Apply the recommended result
111+
thinktank apply
112+
113+
# Apply a specific agent's result
114+
thinktank apply --agent 2
115+
61116
# View the last run's results
62117
thinktank list
63118
```
64119

65-
## Output
120+
## Example output
66121

67122
```
68123
thinktank — ensemble AI coding
@@ -102,9 +157,7 @@ Convergence
102157
| Single Claude Code run | pass@1 | 1x | Fastest |
103158
| **thinktank (N=3)** | **~pass@3** | **3x** | **Same wall time** |
104159
| **thinktank (N=5)** | **~pass@5** | **5x** | **Same wall time** |
105-
| Manual retry loop | pass@k (sequential) | kx | k * slower |
106-
107-
The key insight: **parallel attempts cost more tokens but not more time.** All agents run simultaneously.
160+
| Manual retry loop | pass@k (sequential) | kx | k × slower |
108161

109162
## References
110163

@@ -121,7 +174,3 @@ The key insight: **parallel attempts cost more tokens but not more time.** All a
121174
### Ensemble theory
122175
- *Superforecasting* — Tetlock & Gardner. The aggregate of independent forecasters consistently beats individuals.
123176
- *The Wisdom of Crowds* — Surowiecki. Independent estimates, when aggregated, converge on truth.
124-
125-
## License
126-
127-
MIT

0 commit comments

Comments
 (0)