Add autonomous consolidation loop and react-frontend persona - #9
Merged
Conversation
Introduces the ralph:consolidate skill pair (init + resume) for unattended learnings/skills/guidelines curation in a worktree, with security hooks, two-pass convergence, and autonomous MEDIUM judgment. Adds the react-frontend persona seeded with judgment-layer content distilled from 5 learning files, with conditional references back to the recipe-level details. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…ngs, session retro - Replace type-blocked convergence (2 clean sweeps per type) with round-based (L→S→G per round, 2 consecutive clean rounds to converge). Halves minimum iterations for clean corpora (6 vs 12). - Add compounded-learnings.md for corpus meta-insights, isolated from sweep corpus to prevent feedback loops. Resume surfaces for /learnings:compound. - Add max rounds guard (5 rounds) as circuit breaker. - Add pre-flight cadence analysis to right-size iteration suggestions. - Fix skills glob: */SKILL.md → **/SKILL.md for nested skills. - Preserve notes history (append with ### Iter N, don't overwrite). - Add learnings: skill maturity progression, isolation boundaries, persona judgment-layer framework, round-based vs type-blocked convergence. - Add communication guideline: calibrate challenge intensity to session phase. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
5 tasks
ahoym
added a commit
that referenced
this pull request
May 17, 2026
Addresses team review (#9) on PR #109. Hooks in claude/settings.json fire in every Claude Code session regardless of CWD — adding per-prompt overhead on non-dotfiles machines (Java/frontend/ other codebases). The Rust binary already no-ops without providers.json, but the wrapper still spawns. Add a one-line guard to both wrappers so machines that haven't run setup-claude.sh short-circuit before any binary lookup. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
ahoym
added a commit
that referenced
this pull request
May 18, 2026
Addresses team review (#9) on PR #109. Hooks in claude/settings.json fire in every Claude Code session regardless of CWD — adding per-prompt overhead on non-dotfiles machines (Java/frontend/ other codebases). The Rust binary already no-ops without providers.json, but the wrapper still spawns. Add a one-line guard to both wrappers so machines that haven't run setup-claude.sh short-circuit before any binary lookup. Co-Authored-By: Claude Opus 4.7 (1M context) <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
ralph:consolidateskill pair (init+resume) for unattended learnings/skills/guidelines curation in a worktreereact-frontendpersona seeded with judgment-layer content distilled from 5 learning filesralph/directoryKey changes:
.claude/commands/ralph/consolidate/init/SKILL.md— worktree creation, template scaffolding, pre-flight metrics.claude/commands/ralph/consolidate/resume/SKILL.md— state review, blocker handling, relaunch guidance.claude/ralph/consolidate/— runner script (wiggum.sh), security hooks (bash/web/write-scope guards), output templates, spec.claude/commands/set-persona/react-frontend.md— judgment-layer persona for React/Next.js/Playwright/accessibility with conditional references to recipe-level learnings.claude/learnings/ralph-loop.md— consolidation loop variant docs, stateless spec design patterns.claude/learnings/skill-design.md— three-level skill routing confirmationdocs/plans/consolidation-loop.md— design plan for the consolidation systemScreenshots
N/A
Test plan
/ralph:consolidate:init— worktree created, templates scaffolded, pre-flight populatedwiggum.sh 20— completed 12 iterations, 2 passes, converged cleanly/ralph:consolidate:resume— correctly reported COMPLETE state with 0 actions, 1 LOWsetup-claude.shcreates theralphsymlinkChecklist
Generated with Claude Code