Repository navigation
Log per-call statistics to a JSONL file - #100
Merged
Merged
Conversation
Each run writes one record per agent call (step, category, model, tokens, turns, cost, time) plus run, categories and end records, so cost can be broken down by pipeline step.
- Record a resume's cost as the increase over its session's total - Log schema-validation errors and planSpawn failures as call errors - Log the prompt size the backend actually sends - Write the end record once, before serving and on early return - Always print the call log path - Type every record kind
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Runs are costly, and the end-of-run token table only shows totals per model. This adds a per-run JSONL file so we can see which steps cost the most.
Each run writes to
$TMPDIR/tulip-calls-<timestamp>-<pid>.jsonland logs the path at the end. Records:run: version, agent, PR, diff sizecategories: name, attention, change countscall, one per agent process: step label (e.g.explain,review-amend,mermaid-fix), category, batch/attempt, model, session ids, prompt size, wall time, tokens, turns, cost. Failed calls and schema retries are included.end: success, wall time, errorrunSession/resumeSessionnow require alabel, so every call site names its step.Example:
See docs/adr/0029. Checked with a live run on #25.
claudereports a resume's cost as the session total, so the log records the difference.