this repo has no description
docs: add auth-adapter to architecture diagrams
Update README.md and DEPLOY.md to reflect the current architecture
with auth-adapter between LiteLLM and anthropic-proxy.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
+11
-2
+6
DEPLOY.md
+6
DEPLOY.md
···
390
390
│ │ │
391
391
│ ▼ │
392
392
│ ┌──────────────────┐ │
393
+
│ │ auth-adapter │ │
394
+
│ │ (:4002) headers │ │
395
+
│ └────────┬─────────┘ │
396
+
│ │ │
397
+
│ ▼ │
398
+
│ ┌──────────────────┐ │
393
399
│ │ anthropic-proxy │ │
394
400
│ │ (:4001) OAuth │ │
395
401
│ └────────┬─────────┘ │
+5
-2
README.md
+5
-2
README.md
···
7
7
```
8
8
Telegram Bot (Bun)
9
9
│
10
-
├──► Haiku 4.5 Detection ──► LiteLLM ──► Anthropic API
10
+
├──► Haiku 4.5 Detection ──► LiteLLM ──► auth-adapter ──► anthropic-proxy ──► Anthropic API
11
11
│ (overwhelm, brain dump, self-bullying)
12
12
│
13
13
▼
14
14
Letta (port 8283) - AI agent framework
15
15
↓ OpenAI-compatible API
16
16
LiteLLM (port 4000) - API translation layer
17
-
↓ Anthropic API format (x-api-key via extra_headers)
17
+
↓
18
+
auth-adapter (port 4002) - Bearer → x-api-key header translation
19
+
↓
18
20
anthropic-proxy (port 4001) - OAuth session management
19
21
↓
20
22
Anthropic API (Claude Opus 4.5)
···
24
26
- **Haiku 4.5 Detection**: Fast classification of user messages for overwhelm, brain dumps, and self-bullying
25
27
- **Letta**: AI agent framework with persistent memory (uses Opus 4.5)
26
28
- **LiteLLM**: Translates OpenAI-compatible requests to Anthropic format
29
+
- **auth-adapter**: Translates Bearer tokens to x-api-key headers for anthropic-proxy
27
30
- **anthropic-proxy**: OAuth proxy for Anthropic API access
28
31
- **SQLite**: Local storage for items, wins, and context
29
32