HugeFrog24/go-telegram-bot
1
0
Fork 0
mirror of https://github.com/HugeFrog24/go-telegram-bot.git synced 2026-10-27 21:35:26 +00:00
Code Issues Packages Projects Releases Wiki Activity

Upgrade dependencies and toolchain; migrate to Claude API 2026-10-21

Browse Source 
Toolchain and CI:
- Go 1.27 (go directive 1.27.0, CI 1.27.3)
- actions/checkout v7, actions/setup-go v7
- golangci-lint v2.15.0, gosec v2.30.0
- Dockerfile: drop obsolete -a -installsuffix cgo build flags

Go modules:
- anthropic-sdk-go v1.71.0 -> v2.3.0 (new /v2 module path, API version
  2026-10-21); go-telegram/bot v1.27.0; x/sync v0.24.0; indirects to latest
- BetaMessageParam is now a per-role union: build assistant turns with
  OfAssistant and read content through the role in the cache helper
- Pass the Anthropic API key only when set: v2 sends an explicitly passed
  credential as is, so an empty key became an empty bearer token instead of
  falling back to the SDK's default credential chain as v1 did

API 2026-10-21 / deprecations:
- Stop sending temperature (rejected on every model); the config key is
  still parsed and a startup notice flags it
- Name refusals in the empty-response error instead of reporting an
  unexpected response format
- ElevenLabs STT: scribe_v1 (deprecated) -> scribe_v2
- Template config: claude-haiku-5-5 with effort low, temperature removed

Docs: replace the unsupported `go get -u` install step, document the ignored
temperature key, default-on thinking, and per-model effort defaults.
This commit is contained in:
HugeFrog24
2026-10-25 13:34:39 +01:00
parent ffb3a31db8
commit 401b263991
14 changed files with 210 additions and 81 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+4 -8
View File
@@ -40,12 +40,7 @@ A scalable, multi-bot solution for Telegram using Go, GORM, and the Anthropic AP
### Native Deployment
1. Install using `go get`:
```bash
go get -u github.com/HugeFrog24/go-telegram-bot
cd go-telegram-bot
```
1. Clone the repository as above. Building needs Go 1.27 or later and a C compiler, because the SQLite driver uses cgo.
2. Configure as above, then build:
```bash
@@ -73,10 +68,11 @@ Each bot is one JSON file in `config/` (see `config/default.json` for the templa
| Key | Type | Default | Description |
| ------------------ | ------ | ----------- | ----------- |
| `max_tokens` | number | `1000` | Maximum output tokens per reply. **Thinking tokens count toward this limit** — raise it (e.g. `4000`+) whenever `thinking` is `"adaptive"`, or a turn can spend the whole budget on reasoning and produce no text. |
| `max_tokens` | number | `1000` | Maximum output tokens per reply. **Thinking tokens count toward this limit** — raise it (e.g. `4000`+) whenever `thinking` is `"adaptive"`, or a turn can spend the whole budget on reasoning and produce no text. Claude 5-generation models (Haiku 5.5, Sonnet 5 and later, Opus 5 and later, Fable) think even when `thinking` is omitted; on those, keep `effort` low or raise this limit. |
| `temperature` | number | *(ignored)* | No longer sent: since API version `2026-10-21` the Claude API rejects `temperature`, `top_p` and `top_k` on every model. Old configs still load, and the bot logs a startup notice; remove the key. |
| `thinking` | string | *(omitted)* | Reasoning mode: `"adaptive"` (the model decides when and how much to think) or `"disabled"`. Omit the key entirely to use the model's own API default. If the configured model doesn't support the chosen mode, the API rejects the request with a 400 — owners/admins see the raw error, regular users get the generic fallback. Check [Anthropic's model docs](https://platform.claude.com/docs/en/build-with-claude/adaptive-thinking) for per-model support. |
| `thinking_display` | string | *(omitted)* | `"summarized"` or `"omitted"`. Only valid together with `"thinking": "adaptive"`. Controls whether the API returns a readable summary of the reasoning (logged, never sent to chat). Thinking is billed the same either way; when omitted, the API's per-model default applies. |
| `effort` | string | *(omitted)* | `"low"`, `"medium"`, `"high"`, `"xhigh"` or `"max"`. Caps how much reasoning and token spend a turn may use; the API default is `high`. For short-reply chat bots `"low"` or `"medium"` cuts cost noticeably. Rejected with a 400 on models older than Claude 4.6 (Haiku 4.5, Sonnet 4.5) — omit it there. |
| `effort` | string | *(omitted)* | `"low"`, `"medium"`, `"high"`, `"xhigh"` or `"max"`. Caps how much reasoning and token spend a turn may use; the API default is `high` on most models and `medium` on Haiku 5.5 and Opus 5.5. For short-reply chat bots `"low"` or `"medium"` cuts cost noticeably. Rejected with a 400 on models older than Claude 4.6 (Haiku 4.5, Sonnet 4.5) — omit it there. |
| `debounce_ms` | number | *(omitted)* | Quiet window, in milliseconds, for coalescing rapid follow-up messages into a single turn. Omit or set `0` to disable. See [Coalescing rapid messages](#coalescing-rapid-messages) below. |
| `stream_drafts` | bool | `false` | Streams each reply into one Telegram draft that grows in place, with a native Stop button, and sends it as a single message when done. Private chats only. See [Draft streaming and the Stop button](#draft-streaming-and-the-stop-button) below. |
| `cache_history` | bool | `true` | Places a prompt-cache breakpoint on the trailing conversation block, so each turn reads the prior history from cache instead of reprocessing it at full price. Set `false` to cache only the system prompt. |