Conversation
…eamWithContext) Reasoning models (e.g. Step 3.5 Flash, DeepSeek R1) reject requests when reasoning params are missing. pi-ai's stream() passes options directly to buildParams which checks options.reasoningEffort — only streamSimple() maps options.reasoning → reasoningEffort internally. This adds reasoningEffort: "low" to both chatWithContext and streamWithContext when model.reasoning is true.
Add agent.reasoning_effort config (off/low/medium/high) so users can control thinking depth for reasoning models. Previously hardcoded to "low". - Schema: new enum field in AgentConfigSchema (default: "low") - Backend: chatWithContext + streamWithContext read from config - Model catalog: add reasoning flag to ModelOption for UI hints - WebUI: reasoning effort Select next to model selector, auto-disables (greyed out) when current model does not support reasoning - Select component: add disabled prop support
- /reasoning [off|low|medium|high] — view or set reasoning effort - /model now shows current reasoning effort alongside model name - Help text explains what reasoning levels do and which models support it
cthellla
changed the title
TONresistor
added a commit
that referenced
this pull request
Jun 1, 2026
escapeFts5Query (3 sites) and bm25ToScore (2 sites) were byte-identical clones. Centralise both in memory/search/fts-utils.ts so the FTS5 escape list — a syntax-injection surface — has a single definition. Closes audit finding #13.
cthellla
added a commit
to cthellla/teleton-agent
that referenced
this pull request
Sep 11, 2026
@grammyjs/types goes 3.25.0 -> 5.0.0 with it, which is what actually matters: the typed surface moves from Bot API ~9.5 to 10.3, so sendRichMessage, sendRichMessageDraft, answerGuestQuery, EphemeralMessageParameters, can_stop, DisabledButton, can_send_welcome_messages and poll persistent_id all become callable through grammy instead of raw fetch. That lifts the "grammY has no types yet, go through raw HTTP" blocker shared by TONresistor#13 (streaming stop button), TONresistor#14 (ephemeral group replies) and TONresistor#16 (rich messages in bot DMs), and makes the existing raw-fetch answerGuestQuery in bridges/bot.ts replaceable. grammY tracks Bot API within days (1.42 shipped the day 9.6 landed, 1.46 two days after 10.3), so there is no migration away from it. Despite two major bumps in the types package, no source change was needed: build, typecheck and tests are unchanged (2044 pass, same 6 known failures). Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
Add configurable reasoning effort for thinking models (o3, DeepSeek R1, etc).
Previously reasoning was either unsupported or hardcoded to "low". This PR adds:
agent.reasoning_effortconfig field (off/low/medium/high) with schema, validation, and hot-reload/reasoningadmin command to view/set effort via Telegram/modelnow shows current reasoning effort in statusreasoningflag on known thinking models for UI hints/reasoningregistered insyncCommands()Note: "off" skips sending reasoning params — may not work with reasoning-only models (e.g. DeepSeek R1).
Type of Change
Checklist
mainbranchnpm run typecheckpassesnpm run lintpassesnpm testpasses (1 pre-existing test file fails identically on upstream/main)Related Issues
N/A