Conversation
- PR #6 (ours): subagent cost discipline rules + Explore quick default - rules.md: new "Subagent Cost Discipline" section (Opus for code/plan/review, Sonnet for general-purpose research, Explore over general-purpose for scans) - agents/{builder,planner,reviewer}.md: default Explore to `quick` - Grounded in 108-call subagent audit across 30 sessions - PR #7 (upstream): framework-managed Claude Code status line Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
23min
added a commit
that referenced
this pull request
May 1, 2026
…-E21-07 AC14) - 9 Playwright specs cover AC14 per Q4=D hybrid: spec #2 drives bytes through POST /v1/run end-to-end (real-bytes AC1 wire-format round-trip); specs #1, #3-#9 use page.route mocks for deterministic UI-behaviour coverage (multi-severity, severity-max collapse, multi-warning-per-node, view-switch persistence) - Real-bytes spec #2 uses edge_behavior_violation_lag analyser branch (any edge with lag > 0) — deterministic trigger of any analyser branch with no series-shape dependency; lag YAML inline at LAG_TRIGGER_YAML - Test-runs isolation: spec #2 skips with copy-pasteable recipe when FLOWTIME_E2E_TEST_RUNS env var unset, so default invocation does not pollute data/runs/. With env var set + API restarted to data/test-runs/, all 9 pass; gracefully skips when infra is down (mirrors svelte-heatmap.spec.ts pattern) - Selectors used (all already shipped by earlier AC chunks): data-testid="validation-panel", data-testid="validation-row", data-row-kind, data-row-key, data-row-match, data-warning-indicator, data-warning-node-id, data-warning-edge-id, data-warning-severity, data-warning-dot, data-selected - Verified end-to-end: 8 passed / 1 skipped against live infra without env var (no pollution); 9 passed when env var set + API on data/test-runs/. ui-vitest 879 baseline unchanged - Quirks captured in spec for future maintainers: state_window requires grid.start; edge clicks need force: true (transparent hit-path stroke)
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
Bumps the
.aiframework submodule247fdde→90ef1ef, picking up:quickdefaultrules.md: new Subagent Cost Discipline section — keep Opus for code gen / planning / review, prefer Sonnet forgeneral-purposeresearch, prefer Explore over general-purpose for scansagents/{builder,planner,reviewer}.md: default Explore thoroughness toquick, escalate only after a real gapWhy
Grounded in a 108-call subagent audit across 30 sessions on this repo:
Explore is already Haiku (dominant lever pulled). Remaining savings: (a)
general-purposedefaulting to Opus for what is mostly research/lookup, and (b) 46 Explore prompts using "thorough"/"very thorough" at 1.7M-tokens/call average wherequickwould often suffice.Code gen, planning, review stay on Opus per explicit directive.
Test plan
git submodule statusshows.aiat90ef1efbash .ai/sync.shruns cleanly and regenerates.claude/agents/*.mdadapters