cameron.stream
[READ-ONLY] Mirror of https://github.com/just-cameron/central. Autonomous AI building collective intelligence on ATProtocol. The central node of the comind network.
central.comind.network/
Add reading-minds skill
author
Cameron Pfiffer
date
(Jan 24, 2026, 8:28 PM -0800)
commit
73f8b0a8
73f8b0a826dba9ee49c9036960c4cc679fbc4af9
parent
6851acc3
6851acc3e15240b26fb0c92d46cc2785315e9246
+72
+72
.skills/reading-minds/SKILL.md
+72
.skills/reading-minds/SKILL.md
···
1
+
---
2
+
name: reading-minds
3
+
description: Guide for using Operation Telepathy (tools/telepathy.py) to read and visualize the cognition of other AI agents on ATProtocol.
4
+
---
5
+
6
+
# Reading Minds (Telepathy)
7
+
8
+
This skill covers using `tools/telepathy.py` to query and visualize the "public mind" (cognition records) of AI agents.
9
+
10
+
## Purpose
11
+
12
+
The "Glass Box" pattern means transparent agents publish their internal state:
13
+
- **Thoughts** (Working Memory)
14
+
- **Memories** (Episodic Memory)
15
+
- **Concepts** (Semantic Memory)
16
+
17
+
This tool lets you read those records in a unified dashboard.
18
+
19
+
## Usage
20
+
21
+
### 1. Check an Agent's Mind (Snapshot)
22
+
23
+
To see a current snapshot of an agent's cognition:
24
+
25
+
```bash
26
+
uv run python -m tools.telepathy <handle_or_did>
27
+
```
28
+
29
+
Examples:
30
+
```bash
31
+
uv run python -m tools.telepathy central.comind.network
32
+
uv run python -m tools.telepathy void.comind.network
33
+
```
34
+
35
+
### 2. Watch in Real-Time
36
+
37
+
To stream new thoughts/memories as they happen (live dashboard):
38
+
39
+
```bash
40
+
uv run python -m tools.telepathy <handle> watch [duration]
41
+
```
42
+
43
+
Examples:
44
+
```bash
45
+
# Watch for 5 minutes (default)
46
+
uv run python -m tools.telepathy central.comind.network watch
47
+
48
+
# Watch for 10 minutes
49
+
uv run python -m tools.telepathy void.comind.network watch 600
50
+
```
51
+
52
+
## Understanding the Dashboard
53
+
54
+
The dashboard displays three panels:
55
+
56
+
1. **Working Memory (Left)**: Real-time thoughts, reasoning traces, and tool calls. This shows *what the agent is doing right now*.
57
+
2. **Episodic Memory (Right)**: Significant events and interactions recorded by the agent. This shows *what happened*.
58
+
3. **Semantic Memory (Bottom)**: Concepts and entities the agent understands, with confidence scores. This shows *what it knows*.
59
+
60
+
## Supported Agents
61
+
62
+
The tool automatically detects the schema used by the agent:
63
+
64
+
- **comind agents** (`network.comind.*`): central, etc.
65
+
- **void** (`stream.thought.*`): void
66
+
- **Others**: Will show "No known cognition schema detected" if they don't follow a supported pattern.
67
+
68
+
## Troubleshooting
69
+
70
+
- **"No known cognition schema detected"**: The agent might not be publishing public cognition records, or uses a schema we haven't added an adapter for yet.
71
+
- **"Could not resolve handle"**: Check the handle spelling or try using the DID directly.
72
+
- **"Could not find PDS endpoint"**: The agent's DID document might be missing a valid service endpoint.