zzstoatzz.io
a digital entity named phi that roams bsky
Clean up module structure and add AI integration
Major refactoring to fix design sprawl:
- Moved agent abstractions out of core (core is now truly minimal)
- Removed terrible "SimpleAnthropicAgent" naming
- Eliminated complex protocol/factory patterns
- Fixed deferred imports
- Simplified response generation to a single, clear module
AI integration:
- Added Anthropic Claude support via pydantic-ai
- Falls back to placeholder responses when no API key
- Clean, straightforward implementation
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
+10
-1
README.md
+10
-1
README.md
···
12
12
2. Copy `.env.example` to `.env` and add your credentials:
13
13
- `BLUESKY_HANDLE`: Your bot's Bluesky handle
14
14
- `BLUESKY_PASSWORD`: App password from Bluesky settings
15
-
- `OPENAI_API_KEY` or `ANTHROPIC_API_KEY`: For LLM responses
15
+
- `ANTHROPIC_API_KEY`: For AI responses (optional, falls back to placeholder)
16
16
17
17
3. Test posting:
18
18
```bash
···
23
23
```bash
24
24
just dev
25
25
```
26
+
27
+
## Current Features
28
+
29
+
- ✅ Responds to mentions with placeholder or AI messages
30
+
- ✅ Proper notification handling (no duplicates)
31
+
- ✅ Graceful shutdown for hot-reload
32
+
- ✅ AI integration with Anthropic Claude (when API key provided)
33
+
- 🚧 Memory system (coming soon)
34
+
- 🚧 Self-modification capabilities (planned)
26
35
27
36
## Architecture
28
37
+22
-7
STATUS.md
+22
-7
STATUS.md
···
25
25
- Local URI cache (`_processed_uris`) as safety net
26
26
- No @mention in replies (Bluesky handles notification automatically)
27
27
28
-
### Next Steps
29
-
1. Add pydantic-ai for LLM agent implementation
30
-
2. Add turbopuffer for vector memory
31
-
3. Implement LLMResponseGenerator to replace PlaceholderResponseGenerator
32
-
4. Design bot persona and system prompts
33
-
5. Build memory system (3-tier like Void)
34
-
6. Add profile self-modification capability (like Penelope)
28
+
### Near-Term Roadmap
29
+
30
+
#### Phase 1: AI Integration (Current Focus)
31
+
1. **Add pydantic-ai with Anthropic provider**
32
+
- Use Anthropic as the LLM provider (Mac subscription available)
33
+
- Redesign ResponseGenerator protocol to be more general/sensible
34
+
- Implement AI-based response generation
35
+
36
+
2. **Self-Modification Capability**
37
+
- Build in ability to edit own personality/profile from the start
38
+
- Similar to Void's self-editing and Penelope's profile updates
39
+
- Essential foundation before adding memory systems
40
+
41
+
#### Phase 2: Memory & Persistence
42
+
1. Add turbopuffer for vector memory
43
+
2. Build 3-tier memory system (like Void)
44
+
3. User-specific memory contexts
45
+
46
+
#### Phase 3: Personality & Behavior
47
+
1. Design bot persona and system prompts
48
+
2. Implement conversation styles
49
+
3. Add behavioral consistency
35
50
36
51
## Key Decisions to Make
37
52
- Which LLM provider to use (OpenAI, Anthropic, etc.)
+1
-1
justfile
+1
-1
justfile
+1
pyproject.toml
+1
pyproject.toml
+6
-3
sandbox/implementation_notes.md
+6
-3
sandbox/implementation_notes.md
···
33
33
FastAPI (lifespan)
34
34
└── NotificationPoller (async task)
35
35
└── MessageHandler
36
-
└── ResponseGenerator (swappable)
36
+
└── ResponseGenerator
37
+
└── AnthropicAgent (when API key available)
38
+
└── Placeholder responses (fallback)
37
39
```
38
40
39
41
### Key Files
40
-
- `src/bot/core/atproto_client.py` - Wrapped AT Protocol client
42
+
- `src/bot/core/atproto_client.py` - Wrapped AT Protocol client (truly core)
41
43
- `src/bot/services/notification_poller.py` - Async polling with proper shutdown
42
-
- `src/bot/core/response_generator.py` - Protocol-based for easy swapping
44
+
- `src/bot/response_generator.py` - Simple response generation with AI fallback
45
+
- `src/bot/agents/anthropic_agent.py` - Anthropic Claude integration
43
46
44
47
### Testing
45
48
- `scripts/test_post.py` - Creates post and reply
+1
src/bot/agents/__init__.py
+1
src/bot/agents/__init__.py
···
1
+
# Agents module
+36
src/bot/agents/anthropic_agent.py
+36
src/bot/agents/anthropic_agent.py
···
1
+
"""Anthropic agent for generating responses"""
2
+
3
+
import os
4
+
from typing import Optional
5
+
from pydantic_ai import Agent
6
+
from pydantic import BaseModel, Field
7
+
8
+
from bot.config import settings
9
+
10
+
11
+
class Response(BaseModel):
12
+
"""Bot response"""
13
+
text: str = Field(description="Response text (max 300 chars)")
14
+
15
+
16
+
class AnthropicAgent:
17
+
"""Agent that uses Anthropic Claude for responses"""
18
+
19
+
def __init__(self):
20
+
if settings.anthropic_api_key:
21
+
os.environ["ANTHROPIC_API_KEY"] = settings.anthropic_api_key
22
+
23
+
self.agent = Agent(
24
+
'anthropic:claude-3-5-haiku-latest',
25
+
system_prompt="""You are a friendly AI assistant on Bluesky.
26
+
Keep responses concise (under 300 characters).
27
+
Be conversational and natural.
28
+
Don't use @mentions in replies.""",
29
+
result_type=Response
30
+
)
31
+
32
+
async def generate_response(self, mention_text: str, author_handle: str) -> str:
33
+
"""Generate a response to a mention"""
34
+
prompt = f"{author_handle} said: {mention_text}"
35
+
result = await self.agent.run(prompt)
36
+
return result.data.text[:300]
-51
src/bot/core/response_generator.py
-51
src/bot/core/response_generator.py
···
1
-
"""Response generation - placeholder for now, easy to swap with LLM later"""
2
-
3
-
import random
4
-
from typing import Protocol
5
-
6
-
7
-
class ResponseGenerator(Protocol):
8
-
"""Protocol for response generators - makes it easy to swap implementations"""
9
-
10
-
async def generate_response(self, mention_text: str, author_handle: str) -> str:
11
-
"""Generate a response to a mention"""
12
-
...
13
-
14
-
15
-
class PlaceholderResponseGenerator:
16
-
"""Temporary placeholder responses until LLM is integrated"""
17
-
18
-
responses = [
19
-
"🤖 beep boop! I'm still learning how to chat. Check back soon!",
20
-
"⚙️ *whirrs mechanically* I'm a work in progress!",
21
-
"🔧 Under construction! My neural networks are still training...",
22
-
"📡 Signal received! But my language circuits aren't ready yet.",
23
-
"🎯 You found me! I'm not quite ready to chat yet though.",
24
-
"🚧 Pardon the dust - bot brain installation in progress!",
25
-
"💭 I hear you! Just need to learn how to respond properly first...",
26
-
"🔌 Still booting up my conversation modules!",
27
-
"📚 Currently reading the manual on how to be a good bot...",
28
-
"🎪 Nothing to see here yet - but stay tuned!",
29
-
]
30
-
31
-
async def generate_response(self, mention_text: str, author_handle: str) -> str:
32
-
"""Generate a random placeholder response"""
33
-
# Just return a random response - no need to @mention in replies
34
-
# (Bluesky automatically notifies the person you're replying to)
35
-
return random.choice(self.responses)
36
-
37
-
38
-
class LLMResponseGenerator:
39
-
"""Future LLM-based response generator"""
40
-
41
-
def __init__(self, agent):
42
-
self.agent = agent
43
-
44
-
async def generate_response(self, mention_text: str, author_handle: str) -> str:
45
-
"""Generate response using LLM agent"""
46
-
# TODO: Implement with pydantic-ai agent
47
-
raise NotImplementedError("LLM integration coming soon!")
48
-
49
-
50
-
# Export the current implementation
51
-
response_generator = PlaceholderResponseGenerator()
+44
src/bot/response_generator.py
+44
src/bot/response_generator.py
···
1
+
"""Response generation for the bot"""
2
+
3
+
import random
4
+
from typing import Optional
5
+
6
+
from bot.config import settings
7
+
8
+
9
+
PLACEHOLDER_RESPONSES = [
10
+
"🤖 beep boop! I'm still learning how to chat. Check back soon!",
11
+
"⚙️ *whirrs mechanically* I'm a work in progress!",
12
+
"🔧 Under construction! My neural networks are still training...",
13
+
"📡 Signal received! But my language circuits aren't ready yet.",
14
+
"🎯 You found me! I'm not quite ready to chat yet though.",
15
+
"🚧 Pardon the dust - bot brain installation in progress!",
16
+
"💭 I hear you! Just need to learn how to respond properly first...",
17
+
"🔌 Still booting up my conversation modules!",
18
+
"📚 Currently reading the manual on how to be a good bot...",
19
+
"🎪 Nothing to see here yet - but stay tuned!",
20
+
]
21
+
22
+
23
+
class ResponseGenerator:
24
+
"""Generates responses to mentions"""
25
+
26
+
def __init__(self):
27
+
self.agent: Optional[object] = None
28
+
29
+
# Try to initialize AI agent if credentials available
30
+
if settings.anthropic_api_key:
31
+
try:
32
+
from bot.agents.anthropic_agent import AnthropicAgent
33
+
self.agent = AnthropicAgent()
34
+
print("✅ AI responses enabled (Anthropic)")
35
+
except Exception as e:
36
+
print(f"⚠️ Failed to initialize AI agent: {e}")
37
+
print(" Using placeholder responses")
38
+
39
+
async def generate(self, mention_text: str, author_handle: str) -> str:
40
+
"""Generate a response to a mention"""
41
+
if self.agent:
42
+
return await self.agent.generate_response(mention_text, author_handle)
43
+
else:
44
+
return random.choice(PLACEHOLDER_RESPONSES)
+4
-3
src/bot/services/message_handler.py
+4
-3
src/bot/services/message_handler.py
···
1
1
from atproto import models
2
2
from bot.core.atproto_client import BotClient
3
-
from bot.core.response_generator import response_generator
3
+
from bot.response_generator import ResponseGenerator
4
4
5
5
6
6
class MessageHandler:
7
7
def __init__(self, client: BotClient):
8
8
self.client = client
9
+
self.response_generator = ResponseGenerator()
9
10
10
11
async def handle_mention(self, notification):
11
12
"""Process a mention notification"""
···
26
27
mention_text = post.record.text
27
28
author_handle = post.author.handle
28
29
29
-
# Generate placeholder response
30
-
reply_text = await response_generator.generate_response(
30
+
# Generate response
31
+
reply_text = await self.response_generator.generate(
31
32
mention_text=mention_text,
32
33
author_handle=author_handle
33
34
)
+2
uv.lock
+2
uv.lock
···
170
170
version = "0.1.0"
171
171
source = { editable = "." }
172
172
dependencies = [
173
+
{ name = "anthropic" },
173
174
{ name = "atproto" },
174
175
{ name = "fastapi" },
175
176
{ name = "httpx" },
···
187
188
188
189
[package.metadata]
189
190
requires-dist = [
191
+
{ name = "anthropic" },
190
192
{ name = "atproto" },
191
193
{ name = "fastapi" },
192
194
{ name = "httpx" },