site: show the generated code, invite people to build with Lex - #195
Merged
Merged
Conversation
lex-gpu serves OpenAI chat completions from its own compiled Metal and CUDA kernels. lex-llm drives that shape unchanged, so this is the same nine-module wiring every other backend gets: a lex_gpu_agent() beside vllm_agent() in each agent, an arm in session.lex, a flag in main.lex. It is chat-only today and the README says so plainly. lex-gpu accepts a `tools` list and drops it, so the model is never told the tools exist and its replies carry no tool_calls -- the loop never dispatches, which for a coding agent means it describes work rather than doing it. The wiring lands now so that it starts working the moment lex-gpu renders `tools` into its prompt and splits <think> out of content; neither needs a change on this side. Depends on alpibrusl/lex-llm#64 for providers.lex_gpu_model and lex_gpu_local, so CI fails here until that merges. Checked against that branch locally: all nine agents, session.lex and main.lex type-check with no errors, `lex fmt --check` is clean across the fourteen touched files, and `lex doc-sync --check` reports all targets current. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The no-args help banner listed every provider flag except the one just added. README said "nine provider backends" for a list of 8 named + 2 in the table = 10 (already undercounting by one before this branch). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The typed-issue demo ended in an opaque hash and "verified" — proof without anything to look at. Re-recorded it live against this repo (same issue, real Ollama run, ~34 real minutes this time since the model second-guessed a generic type it had already gotten right) with one more real step: print the function it actually wrote. Also softened "a language with no users but us" (said twice — once in the meta description, once in the hero) so the honest point about Lex's current adoption doesn't read as a closed door. Lex is meant to be built with, not just by us. Co-Authored-By: Claude Sonnet 5 <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.
Summary
verified— proof without anything to look at. Added a real fourth step (awk '/^fn zip/,/^}/' src/list_lex.lex) that prints the actual function the run wrote, both in the static snippet and in the recording.site/demo/issue.castlive against this repo (same issue, real Ollama/qwen3.8 run) rather than editing the existing one, then appended the new step's own real recording, preserving theidle_time_limit: 2.0convention the other cast uses.Test plan
lex issue create→lex-code --issue=$ISSUE_ID --ollama→ verified →awkprints the realfn zipbody.castfile decodes to the exact expected terminal transcriptpagesCI on this PR🤖 Generated with Claude Code