Conversation
CONTENT_DRIFT diagnosis was removed as part of the Translation Validation philosophy shift. Clean up the remaining workflow text in both Chinese and English prompt templates.
Multi-tenant e-commerce SaaS benchmark with: - Complete OpenSpec (2864 lines, 20 rounds, Chinese) - OpenAPI 3.1 contract (45 endpoints, 36 schemas) - 287 data-driven E2E test cases across 20 rounds - Test runner with variable capture and cross-round persistence - /bench-ai and /bench-svp slash commands for execution - CONTENT_DRIFT dead documentation cleanup in shared.ts
Instead of reimplementing SVP methodology, bench-svp now instructs the agent to strictly follow the original /svp skill. Only change: alignment checkpoints auto-approve instead of waiting for user confirmation. Key enforcements added: - Must load and follow .claude/commands/svp.md - Forbidden to skip SVP flow or write L1 code directly - Every new endpoint must have a corresponding L3 block - All code changes via subagent dispatch only
Teaches AI agents the WHY behind SVP, not just the HOW: - L3 IS the API contract (pins = params, constraints = rules) - One-way compilation: errors trace upward, never edit L1 directly - Reference docs belong in refs/ (auto-injected into compile prompts) - Context isolation: main Agent orchestrates, subagents write code This addresses the root cause of benchmark failures: agents abandoned SVP methodology because they understood the steps but not the principles. When compilation output didn't match the API contract, they defaulted to directly editing L1 instead of fixing L3 and recompiling.
There was a problem hiding this comment.
Pull request overview
This PR updates the generated /svp skill content (shared across host adapters) by adding a new “SVP Philosophy” section, removes dead CONTENT_DRIFT workflow documentation from the shared workflow templates, and tweaks repo housekeeping around benchmarks/ignores.
Changes:
- Add a Philosophy section (EN/ZH) to the shared skill file assembly.
- Remove
CONTENT_DRIFTblocks from the shared workflow templates. - Adjust TypeScript/glob and ignore configuration as part of benchmark cleanup.
Reviewed changes
Copilot reviewed 1 out of 3 changed files in this pull request and generated no comments.
| File | Description |
|---|---|
| tsconfig.json | Updates TS include globs (currently includes benchmark/**/*.ts). |
| packages/skills/adapters/shared.ts | Adds getPhilosophySection() and injects it into skill file generation; removes CONTENT_DRIFT workflow documentation blocks. |
| .gitignore | Comment-only clarification for hidden-directory re-ignore/whitelist behavior. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
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
/svpskill (all adapters)Key Changes
SVP Philosophy in /svp Skill
Teaches AI agents the WHY behind SVP methodology:
Cleanup
Test plan
npx tsc --noEmit— zero errorsnpx vitest run— 730/730 tests pass