Redesign/controlflow claude code standalone#6
Merged
Smithbox-ai merged 11 commits intoJun 24, 2026
Conversation
Co-Authored-By: Claude <noreply@anthropic.com>
Co-Authored-By: Claude <noreply@anthropic.com>
…model Co-Authored-By: Claude <noreply@anthropic.com>
Co-Authored-By: Claude <noreply@anthropic.com>
…odel Reframe the bilingual tutorial (en + ru, 20 chapters each) and the agent-engineering policy docs around the slim surface: one @controlflow-planner agent, three skills (controlflow-plan / -verify / -review), and a routing stub over native Copilot. The 8 executor roles and 3 inline verify roles are preserved as conceptual labels, not shipped agents; the Orchestrator conductor, dispatch state machine, waves, gates, and the model-routing / tool-grants / agent-grants surfaces are presented only as retired history. Engineering docs reindexed (README), reframed (MIGRATION-CORE-FIRST, PART-SPEC, CLARIFICATION-POLICY, AGENT-AS-TOOL, FINAL-REVIEW-SCOPE, CURSOR-SUPPORT, CURSOR-SDK, RISK-TAXONOMY), and updated with eval-coupled anchors preserved (MEMORY-ARCHITECTURE Memory Content Taxonomy + content_types; PROMPT-BEHAVIOR-CONTRACT Memory Use Discipline invariants). ADR-0001 plan-review handoff updated from the retired Orchestrator to controlflow-verify. Gate green: cd evals && npm test exits 0. No live stale references to the 13-agent model remain in docs; remaining Orchestrator mentions are explicitly retired/negation framing or historical motivation records. Co-Authored-By: Claude <noreply@anthropic.com>
…irst model Rework plugins/controlflow-shared-source/ as the slim sync-OUT source: its skills/ tree now carries the three canonical skills (controlflow-plan, controlflow-verify, controlflow-review) sourced from .github/skills/, the generation manifest declares only those three, and host-overrides carry the per-host invocation prefix plus the Cursor planner-agent overlay. The codex and cursor plugin packages are regenerated from it: the nine obsolete heavy-model skills (orchestration, router, spec, strict-workflow, plan-audit, planning, assumption-verifier, executability-verifier, memory-hygiene), their per-skill agents, retired report templates, and heavy-model validate scripts are removed; both plugin manifests move to 1.0.0 with slim descriptions and the Cursor agents/ directory ships the @controlflow-planner agent. The claude-code plugin is bumped to 1.0.0 and its changelog records the slim surface. Eval tests that hardcoded the old skill names (skill-discoverability, ponytail-adaptation) are repointed at the slim skills; the parity suite (codex == shared-source, no delta) and the core-portability-matrix (Pass 16) now cover the slim surface. Gate green: rm -rf .cache && npm test exits 0. Plugin generation parity VALID (28 declared files). No live stale references to the 13-agent model remain under plugins/; remaining old-skill mentions are retired negation framing or changelog history. Co-Authored-By: Claude <noreply@anthropic.com>
Bump the Claude Code plugin and its local marketplace entry to 1.0.0 (the plugins' first stable release; previously 0.x), update the README plugin section to 1.0.0 with the slim multi-host surface description, and add the v2.0.0 changelog entry for the core breaking refactor (13 -> 1 agent, 3 skills, 0 subagents, native Copilot delegation, retired Orchestrator / model-routing / tool-grants / agent-grants, rewritten evals and bilingual docs, plugins synced from canonical). The core repo version is 2.0.0 (breaking change after 1.3.0); the plugin packages reach their own first stable 1.0.0. Gate green: rm -rf .cache && npm test exits 0. Co-Authored-By: Claude <noreply@anthropic.com>
The slim refactor retired the Orchestrator agent but left the plan-format schema machine-requiring handoff.target_agent: "Orchestrator" and the plan template instructing authors to write it — the shipped contract still named a retired agent. Repoint the const to controlflow-verify (the artifact-first review handoff), refresh the template's Handoff / Execution Notes / Executor Agent / Revision Policy framing to the slim model, move the six planner fixtures and two behavior scenarios to the slim handoff vocabulary (orchestrator_review_applies -> verify_review_applies), and drop the stale heavy-model [Unreleased] changelog block that contradicted the 2.0.0 release. Evals green: cd evals && rm -rf .cache && npm test (EXIT 0, all suites pass). Co-Authored-By: Claude <noreply@anthropic.com>
extractLevel2Headings split on '\n', so CRLF tutorial chapters left a trailing '\r' on each line; '^##\s+(.+)$' then failed to match (JS '.' / '$' exclude '\r'), yielding zero headings on both sides and a vacuous pass. Split on /\r?\n/ so CRLF chapters are actually checked. Co-Authored-By: Claude <noreply@anthropic.com>
Co-Authored-By: Claude <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.
What does this PR do?
Brief description of the change.
Related Issue
Closes #
Type of Change
Verification
cd evals && npm testpasses (full offline suite)npm run test:structuralpassesnpm run test:behaviorpassesChecklist
governance/tool-grants.jsonif agent tools changedgovernance/agent-grants.jsonif delegation changedplans/project-context.mdupdated if agent roster changedskills/index.mdupdated if skill added/modified