agent: spend guardrails — sonnet default, usage report, token budget - #233
Merged
Conversation
… budget The BYO-key agent's Anthropic path defaulted to claude-opus-4-8 with no usage visibility, so live loops silently billed Opus rates per turn (each turn ships 22 tool schemas + the growing conversation + VM output). Three guardrails: - default model claude-opus-4-8 -> claude-sonnet-5 (near-Opus agentic quality at 3/15 vs 5/25 per MTok); Opus/Fable remain one flag away via --model / QOS_CLAUDE_MODEL - per-run usage accounting: input/output/cache tokens accumulated per turn and printed at exit with a sticker-price spend estimate - --token-budget (default 400k, env QOS_TOKEN_BUDGET, 0 = unlimited): cumulative token cap that stops the loop mid-run, complementing the --max-turns cap which bounds turns but not tokens-per-turn Co-Authored-By: Claude Fable 5 <[email protected]>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fixes the uncontrolled API spend from live agent runs (2026-07-20 drain): the Anthropic path defaulted to claude-opus-4-8 with zero usage visibility and no cumulative cap.
--model/QOS_CLAUDE_MODEL).--token-budget(default 400k,QOS_TOKEN_BUDGET, 0 = unlimited): cumulative token cap that halts the loop mid-run.--max-turnsbounds turns but not tokens-per-turn; this bounds the actual spend.🤖 Generated with Claude Code