@askalf/dario 2.9.0 → 2.9.1
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 +8 -7
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -68,7 +68,7 @@ Opus, Sonnet, Haiku — all models, streaming, tool use. Works with Cursor, Cont
|
|
|
68
68
|
<tr>
|
|
69
69
|
<td colspan="3" valign="top">
|
|
70
70
|
|
|
71
|
-
*"The 429s were driving us crazy running a multi-agent stack on Claude Max. You found the billing tag, fixed the checksum, reverse-engineered the per-request hash from the binary —
|
|
71
|
+
*"The 429s were driving us crazy running a multi-agent stack on Claude Max. You found the billing tag, fixed the checksum, reverse-engineered the per-request hash from the binary — running clean, zero reclassification."* — [@belangertrading](https://github.com/belangertrading), multi-agent stack on Claude Max
|
|
72
72
|
|
|
73
73
|
</td>
|
|
74
74
|
</tr>
|
|
@@ -87,7 +87,7 @@ dario is the only proxy that solves this. It injects native Claude Code device i
|
|
|
87
87
|
| **Billing classification** | Native Claude Code session | Third-party (Extra Usage) |
|
|
88
88
|
| **Max plan limits** | Used correctly | Bypassed — billed separately |
|
|
89
89
|
| **Device identity** | Injected automatically | Missing |
|
|
90
|
-
| **Priority routing** |
|
|
90
|
+
| **Priority routing** | Full billing tag fingerprint | Missing |
|
|
91
91
|
| **Billing tag fingerprint** | Per-request SHA-256 matching binary RE | Static or missing |
|
|
92
92
|
| **Beta flags** | Match Claude Code v2.1.100 | Outdated or missing |
|
|
93
93
|
| **Billable beta filtering** | Strips surprise charges | Passes everything through |
|
|
@@ -215,7 +215,7 @@ Model: claude-opus-4-6 (all requests)
|
|
|
215
215
|
|
|
216
216
|
## Passthrough Mode
|
|
217
217
|
|
|
218
|
-
For tools
|
|
218
|
+
For tools that need exact Anthropic protocol fidelity with zero modification, use `--passthrough`. This does OAuth swap only — no billing tag, no thinking injection, no device identity, no extra beta flags. Note: most tools (including Hermes and OpenClaw) work better through default mode, which handles billing classification and token optimization automatically.
|
|
219
219
|
|
|
220
220
|
```bash
|
|
221
221
|
dario proxy --passthrough # Thin proxy, zero injection
|
|
@@ -417,9 +417,9 @@ Then run `hermes` normally — it routes through dario using your Claude subscri
|
|
|
417
417
|
### Direct API Mode
|
|
418
418
|
- All Claude models (Opus 4.6, Sonnet 4.6, Haiku 4.5) + 1M extended context aliases (`opus1m`, `sonnet1m`)
|
|
419
419
|
- **Native billing classification** — device identity, per-request billing tag with SHA-256 checksums matching real Claude Code (extracted via binary RE), ensures Max plan limits work correctly
|
|
420
|
-
- **
|
|
420
|
+
- **Stealth layer** (v2.9.0) — strips thinking blocks from conversation history (saves 50-80% input tokens), scrubs non-CC fields (`temperature`, `top_p`, `top_k`, `stop_sequences`, `service_tier`), reorders JSON fields to match Claude Code's exact field order, and normalizes system prompts to exactly 3 blocks. Every request is indistinguishable from real Claude Code traffic.
|
|
421
421
|
- **Adaptive thinking** — matches Claude Code's `{ type: 'adaptive' }` mode for optimal reasoning (auto-skipped for Haiku 4.5)
|
|
422
|
-
- **Effort control** — injects `output_config: { effort: '
|
|
422
|
+
- **Effort control** — injects `output_config: { effort: 'medium' }` matching Claude Code's default, or passes through client-specified effort level
|
|
423
423
|
- **Enriched 429 errors** — rate limit errors include utilization %, limiting window, and reset time instead of Anthropic's default `"Error"` message
|
|
424
424
|
- **Auto CLI fallback** — if the API returns 429 and Claude Code is installed, transparently retries through `claude --print` with SSE conversion
|
|
425
425
|
- **OpenAI-compatible** (`/v1/chat/completions`) — works with any OpenAI SDK or tool
|
|
@@ -428,6 +428,7 @@ Then run `hermes` normally — it routes through dario using your Claude subscri
|
|
|
428
428
|
- System prompts and multi-turn conversations
|
|
429
429
|
- Prompt caching and extended thinking
|
|
430
430
|
- **Billable beta filtering** — strips `extended-cache-ttl` from client betas (the only prefix requiring Extra Usage)
|
|
431
|
+
- **Beta deduplication** — client-provided betas are deduplicated against the base set before appending
|
|
431
432
|
- **Orchestration tag sanitization** — strips agent-injected XML (`<system-reminder>`, `<env>`, `<task_metadata>`, etc.) before forwarding
|
|
432
433
|
- **Token anomaly detection** — warns on context spike (>60% input growth) or output explosion (>2x previous)
|
|
433
434
|
- Concurrency control (max 10 concurrent upstream requests)
|
|
@@ -442,9 +443,9 @@ Then run `hermes` normally — it routes through dario using your Claude subscri
|
|
|
442
443
|
|
|
443
444
|
### Passthrough Mode
|
|
444
445
|
- All Claude models with native streaming and tool use
|
|
445
|
-
- OAuth token swap only — no billing tag, thinking, effort,
|
|
446
|
+
- OAuth token swap only — no billing tag, thinking, effort, or device identity injection
|
|
446
447
|
- Minimal beta flags (`oauth-2025-04-20` + client betas only)
|
|
447
|
-
- For tools
|
|
448
|
+
- For tools that need exact Anthropic protocol fidelity with zero modification
|
|
448
449
|
|
|
449
450
|
## Endpoints
|
|
450
451
|
|