Repository navigation
Conversation
Replace the tracker ladder with the current one: convenience functions, config() with handler packages, custom handlers, and manual track() for feedback. Rewrite each provider reference around handlers, document what the SDK records and what it doesn't (time to first token, feedback), add a legacy-to-current mapping, and fix the metrics API timestamps, which are milliseconds.
…lers
Custom handlers now fill {{name}} placeholders with parse_template and drop the tools key from model.parameters before calling the provider. Note that the Python OpenAI handler doesn't forward parameters, that inspect_config is async, and that the to_lang_graph runner is called with invoke().
…icModel anthropic 2.x removed both arguments from messages.create, so forwarding them from the variation raises TypeError.
Skill eval results
Only suites whose source actually changed since their last recorded score were re-run. Soft-failing while we stabilise the baseline. |
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
config()+ handler package → custom handler → manualtrack()for feedback.{{name}}withparse_templateand droptoolsfrommodel.parameters; Strands dropstemperature/top_pfor anthropic 2.x.metrics-api.mdto use milliseconds.Jira: DEVREL-1645
Testing
Ran each reference's code in the cookbook notebooks (OpenAI, Anthropic, Bedrock, Gemini, LangChain, LangGraph, Strands, streaming, feedback, metrics API). Verified feedback events through the metrics API.
Notes
launchdarkly-ai-server0.3.0 (Python) and@launchdarkly/ai-node0.3.0 (Node.js).skills.json, so later merges may needpython3 scripts/generate_catalog.pyre-run.Note
Overview
Rolls the
built-in-metricsskill to v2 forlaunchdarkly-ai-server/@launchdarkly/ai-node0.3.0+, replacing the old tracker/trackMetricsOf/ManagedModelguidance with a four-step ladder: convenience functions →config()+ handler packages → custom handlers → manualtrack()for feedback only.The main skill now states that metrics are automatic when calls go through
invoke()/stream(), documents recorded$ld:ai:*events (including tool calls), clarifies that TTFT is not emitted, adds a legacy-to-current mapping table, and updates guardrails (shutdown(), disabled-config behavior, no double-tracking).Per-provider reference docs (OpenAI, Anthropic, Bedrock, Gemini, LangChain/LangGraph, Strands, streaming) are rewritten around handler package names,
create_handlerpatterns,parse_template, and droppingtoolsfrommodel.parameters. Streaming drops manual TTFT wiring in favor ofstream().skills.jsonpicks up the new description, compatibility line, and 2.0.0-experimental version.metrics-api.mdfixes queryfrom/toto Unix milliseconds and notes TTFT stays zero on the new SDKs.Reviewed by Cursor Bugbot for commit b072236. Bugbot is set up for automated code reviews on this repo. Configure here.