feat: cost-aware harness model eval suite#914
Merged
simple-agent-manager[bot] merged 6 commits intomainfrom May 6, 2026
Merged
feat: cost-aware harness model eval suite#914simple-agent-manager[bot] merged 6 commits intomainfrom
simple-agent-manager[bot] merged 6 commits intomainfrom
Conversation
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
6 scenarios (weather baseline + 5 coding tasks), 3 models
(Gemma 4 26B, GPT-4.1 Mini, Claude Haiku 4.5), cost-per-success
scoring, full JSON trace persistence.
Scenarios: read-and-summarize, grep-locate-code, missing-file-recovery,
propose-patch, interpret-test-failure, weather-baseline.
All scenarios use virtual filesystems for deterministic execution.
Routes through SAM's single AI Gateway ("sam").
Workers AI treated as Cloudflare-billed, not free.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Fix Gemma model ID: gemma-4-27b-it → gemma-4-26b-a4b-it - Fix env var name: WORKERS_AI_COST_PER_1K → WORKERS_AI_COST_PER_1K_TOKENS - Rewrite trace schema to match actual EvalTrace/ScenarioResult types - Fix cost.ts comment about PLATFORM_AI_MODELS import Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
experiments/harness-eval/with 6 scenarios (1 baseline + 5 coding), 3 models (Gemma 4 26B, GPT-4.1 Mini, Claude Haiku 4.5), cost-per-success scoring, and full JSON trace persistenceValidation
pnpm lintpnpm typecheckpnpm testtsc --noEmiton experiment tsconfigStaging Verification (REQUIRED for all code changes — merge-blocking)
N/A: docs-only — this PR only adds files under
experiments/harness-eval/. No deployed code is changed.End-to-End Verification (Required for multi-component changes)
N/A: standalone experiment, no multi-component interaction.
Post-Mortem (Required for bug fix PRs)
N/A: not a bug fix.
Agent Preflight (Required)
Classification
External References
experiments/ai-gateway-tool-call/experiment.tsandFINDINGS-gemma.mdpackages/shared/src/constants/ai-services.ts(lines 100-399)docs/architecture/agent-harness-integration.mdpackages/harness/agent/loop.go,packages/harness/llm/types.goCodebase Impact Analysis
Only
experiments/harness-eval/— no changes to deployed packages (apps/,packages/).Documentation & Specs
experiments/harness-eval/README.md— comprehensive docs for running, interpreting, and extending the suiteConstitution & Risk Check
Principle XI (No Hardcoded Values): Workers AI cost is configurable via
WORKERS_AI_COST_PER_1K_TOKENSenv var. Gateway ID configurable viaAI_GATEWAY_ID. Scenarios and models filterable via env vars. No security risk — experiment does not modify any deployed code.Specialist Review Evidence
🤖 Generated with Claude Code