feat(nix): migrate the devcontainer to a Nix toolchain + Claude-native setup (#625)

assets/workspace/.cursor/agent-models.toml → .claude/agent-models.toml

@@ -1,5 +1,5 @@
-# .cursor/agent-models.toml
-# Single source of truth for cursor-agent model assignments.
+# .claude/agent-models.toml
+# Single source of truth for agent model assignments.
 # Referenced by: justfile.worktree (worktree-start recipe)
 
 [models]

.claude/commands/ci_check.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/ci_check/SKILL.md`.
+Read and follow the workflow in `.claude/skills/ci_check/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/ci_fix.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/ci_fix/SKILL.md`.
+Read and follow the workflow in `.claude/skills/ci_fix/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/code_debug.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/code_debug/SKILL.md`.
+Read and follow the workflow in `.claude/skills/code_debug/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/code_execute.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/code_execute/SKILL.md`.
+Read and follow the workflow in `.claude/skills/code_execute/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/code_review.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/code_review/SKILL.md`.
+Read and follow the workflow in `.claude/skills/code_review/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/code_tdd.md

@@ -1,4 +1,4 @@
-Read and follow the workflow in `.cursor/skills/code_tdd/SKILL.md`.
-Also read `.cursor/rules/tdd.mdc` for the scenario checklist and TDD rules.
+Read and follow the workflow in `.claude/skills/code_tdd/SKILL.md`.
+Also read `.claude/skills/tdd/SKILL.md` for the scenario checklist and TDD rules.
 
 Context: $ARGUMENTS

.claude/commands/code_verify.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/code_verify/SKILL.md`.
+Read and follow the workflow in `.claude/skills/code_verify/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/design_brainstorm.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/design_brainstorm/SKILL.md`.
+Read and follow the workflow in `.claude/skills/design_brainstorm/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/design_plan.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/design_plan/SKILL.md`.
+Read and follow the workflow in `.claude/skills/design_plan/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/git_commit.md

@@ -1,4 +1,4 @@
-Read and follow the workflow in `.cursor/skills/git_commit/SKILL.md`.
-Also read `.cursor/rules/commit-messages.mdc` for the commit message format.
+Read and follow the workflow in `.claude/skills/git_commit/SKILL.md`.
+Also read the Commit Message Standard in `CLAUDE.md` for the commit message format.
 
 Context: $ARGUMENTS

.claude/commands/inception_architect.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/inception_architect/SKILL.md`.
+Read and follow the workflow in `.claude/skills/inception_architect/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/inception_explore.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/inception_explore/SKILL.md`.
+Read and follow the workflow in `.claude/skills/inception_explore/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/inception_plan.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/inception_plan/SKILL.md`.
+Read and follow the workflow in `.claude/skills/inception_plan/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/inception_scope.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/inception_scope/SKILL.md`.
+Read and follow the workflow in `.claude/skills/inception_scope/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/issue_claim.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/issue_claim/SKILL.md`.
+Read and follow the workflow in `.claude/skills/issue_claim/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/issue_create.md

@@ -1,4 +1,4 @@
-Read and follow the workflow in `.cursor/skills/issue_create/SKILL.md`.
+Read and follow the workflow in `.claude/skills/issue_create/SKILL.md`.
 Also read `.github/ISSUE_TEMPLATE/` and `.github/label-taxonomy.toml` for templates and labels.
 
 Context: $ARGUMENTS

.claude/commands/issue_triage.md

@@ -1,4 +1,4 @@
-Read and follow the workflow in `.cursor/skills/issue_triage/SKILL.md`.
+Read and follow the workflow in `.claude/skills/issue_triage/SKILL.md`.
 Also read `.github/label-taxonomy.toml` for canonical labels.
 
 Context: $ARGUMENTS

.claude/commands/pr_create.md

@@ -1,4 +1,4 @@
-Read and follow the workflow in `.cursor/skills/pr_create/SKILL.md`.
+Read and follow the workflow in `.claude/skills/pr_create/SKILL.md`.
 Also read `.github/pull_request_template.md` for the PR template.
 
 Context: $ARGUMENTS

.claude/commands/pr_post-merge.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/pr_post-merge/SKILL.md`.
+Read and follow the workflow in `.claude/skills/pr_post-merge/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/pr_solve.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/pr_solve/SKILL.md`.
+Read and follow the workflow in `.claude/skills/pr_solve/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/worktree_ask.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/worktree_ask/SKILL.md`.
+Read and follow the workflow in `.claude/skills/worktree_ask/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/worktree_brainstorm.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/worktree_brainstorm/SKILL.md`.
+Read and follow the workflow in `.claude/skills/worktree_brainstorm/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/worktree_ci-check.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/worktree_ci-check/SKILL.md`.
+Read and follow the workflow in `.claude/skills/worktree_ci-check/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/worktree_ci-fix.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/worktree_ci-fix/SKILL.md`.
+Read and follow the workflow in `.claude/skills/worktree_ci-fix/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/worktree_execute.md

@@ -1,4 +1,4 @@
-Read and follow the workflow in `.cursor/skills/worktree_execute/SKILL.md`.
-Also read `.cursor/rules/tdd.mdc` for TDD rules.
+Read and follow the workflow in `.claude/skills/worktree_execute/SKILL.md`.
+Also read `.claude/skills/tdd/SKILL.md` for TDD rules.
 
 Context: $ARGUMENTS

.claude/commands/worktree_plan.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/worktree_plan/SKILL.md`.
+Read and follow the workflow in `.claude/skills/worktree_plan/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/worktree_pr.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/worktree_pr/SKILL.md`.
+Read and follow the workflow in `.claude/skills/worktree_pr/SKILL.md`.
 
 Context: $ARGUMENTS

.claude/commands/worktree_solve-and-pr.md

@@ -1,4 +1,4 @@
-Read and follow the workflow in `.cursor/skills/worktree_solve-and-pr/SKILL.md`.
-Also read `.cursor/rules/subagent-delegation.mdc` for delegation patterns.
+Read and follow the workflow in `.claude/skills/worktree_solve-and-pr/SKILL.md`.
+Also read `.claude/skills/subagent-delegation/SKILL.md` for delegation patterns.
 
 Context: $ARGUMENTS

.claude/commands/worktree_verify.md

@@ -1,3 +1,3 @@
-Read and follow the workflow in `.cursor/skills/worktree_verify/SKILL.md`.
+Read and follow the workflow in `.claude/skills/worktree_verify/SKILL.md`.
 
 Context: $ARGUMENTS

assets/workspace/.cursor/rules/branch-naming.mdc → .claude/skills/branch-naming/SKILL.md

@@ -1,6 +1,7 @@
 ---
-description: Topic branch naming and workflow for starting work on an issue. Attach when creating branches, starting work on issues, or checking out branches.
-alwaysApply: false
+name: branch-naming
+description: Topic branch naming and workflow for starting work on an issue. Use when creating branches, starting work on issues, or checking out branches.
+disable-model-invocation: true
 ---
 
 # Topic Branch Naming and Workflow
@@ -40,15 +41,19 @@ When the user asks to create or start work on an issue (e.g. "create branch for
 ## Branch name format (reference)
 
 ### Issue-tied branches
+
 ```
 <type>/<issue_number>-<short_summary>
 ```
+
 Example: `feature/36-standardize-commit-messages`, `bugfix/42-fix-login-bug`
 
 ### Chore branches (no issue required)
+
 ```
 chore/<short_summary>
 ```
+
 Example: `chore/sync-main-to-dev`, `chore/update-dependencies`
 
 ## Branch types (reference)

.cursor/skills/code_review/SKILL.md → .claude/skills/code_review/SKILL.md

@@ -65,7 +65,7 @@ STEPS:
    Flag any change NOT traceable to a requirement (scope creep).
 4. Check project standards:
    - Changelog: is CHANGELOG.md updated under ## Unreleased? Does the entry match?
-   - Commit messages: do all commits follow the format in .cursor/rules/commit-messages.mdc?
+   - Commit messages: do all commits follow the format in CLAUDE.md (Commit Message Standard)?
    - Tests: are there tests for new/changed behavior?
    - Docs: are documentation changes needed?
 5. Produce your report in EXACTLY this structure:

.cursor/skills/inception_explore/README.md → .claude/skills/inception_explore/README.md

@@ -180,4 +180,4 @@ User: "Add multi-tenancy to the system"
 - [RFC template](../../templates/RFC.md)
 - [DESIGN template](../../templates/DESIGN.md)
 - [Keep a Changelog](https://keepachangelog.com/) — format for CHANGELOG.md entries
-- [Single Source of Truth rule](../../../.cursor/rules/single-source-of-truth.mdc)
+- [Single Source of Truth rule](../../../CLAUDE.md)

.cursor/skills/issue_create/SKILL.md → .claude/skills/issue_create/SKILL.md

@@ -33,7 +33,7 @@ Create a new GitHub issue using the appropriate issue template.
    - Draft the body with all required fields from the chosen template.
    - Include a Changelog Category value based on the issue type.
    - For testable issue types (`feature`, `bug`, `refactor`), include a TDD acceptance criterion:
-     `- [ ] TDD compliance (see .cursor/rules/tdd.mdc)`
+     `- [ ] TDD compliance (see .claude/skills/tdd/SKILL.md)`
 
 4. **Show draft and ask for confirmation**
    - Present the title, labels, and body to the user.

.cursor/skills/solve-and-pr/SKILL.md → .claude/skills/solve-and-pr/SKILL.md

@@ -29,7 +29,7 @@ This command:
 - Resolves or creates the linked branch
 - Sets up the environment (`uv sync`, `pre-commit install`)
 - Captures the local gh user as the reviewer (`gh api user --jq '.login'`)
-- Launches a tmux session running `cursor-agent` with `--yolo` mode
+- Launches a tmux session running `claude --dangerously-skip-permissions`
 - Passes `/worktree-solve-and-pr` as the initial prompt
 
 ### 3. Report back to the user

.cursor/rules/subagent-delegation.mdc → .claude/skills/subagent-delegation/SKILL.md

@@ -1,10 +1,16 @@
+---
+name: subagent-delegation
+description: How to delegate mechanical sub-steps to lightweight subagents when executing skills. Use when running a skill that has data-gathering, formatting, or structured-review sub-steps.
+disable-model-invocation: true
+---
+
 # Subagent Delegation
 
 When executing skills, delegate mechanical sub-steps to lightweight subagents via the Task tool to reduce token consumption on the primary model.
 
 ## Model Tiers
 
-See [.cursor/agent-models.toml](../.cursor/agent-models.toml) for the single source of truth. Summary:
+See [.claude/agent-models.toml](../../agent-models.toml) for the single source of truth. Summary:
 
 - **lightweight** (`composer-1.5`) — CLI commands, API calls, file reading, parsing, template filling
 - **standard** (`sonnet-4.5`) — structured analysis, code review with clear inputs
@@ -84,7 +90,7 @@ The following steps SHOULD be delegated to reduce token consumption:
 - **Step 6** (publish comment): Spawn a Task subagent with `model: "fast"` that
   posts the formatted comment and returns the comment URL.
 
-Reference: [subagent-delegation rule](../../rules/subagent-delegation.mdc)
+Reference: [subagent-delegation skill](../subagent-delegation/SKILL.md)
 ```
 
 ## Important Notes

.cursor/rules/tdd.mdc → .claude/skills/tdd/SKILL.md

@@ -1,26 +1,19 @@
 ---
-description: TDD discipline and test scenario guidance when writing code
-alwaysApply: false
-globs:
-  - "**/*.py"
-  - "**/*.ts"
-  - "**/*.js"
-  - "**/*.sh"
-  - "**/test_*"
-  - "**/*_test.*"
-  - "**/tests/**"
+name: tdd
+description: TDD discipline and test scenario guidance when writing code. Use when implementing features or fixes that have testable behavior.
+disable-model-invocation: true
 ---
 
 # TDD
 
 When implementing features or fixes that have testable behavior:
 
 1. Write the failing test first. Run it. Confirm it fails.
-2. **Commit** the failing test following [commit-messages.mdc](commit-messages.mdc) (`test: ...`). Do not proceed before committing.
+2. **Commit** the failing test following the commit message standard in `CLAUDE.md` (`test: ...`). Do not proceed before committing.
 3. Write minimal code to make the test pass. Run it. Confirm it passes. **Commit** the implementation.
 4. Refactor. Run tests. Confirm no regressions. **Commit** the refactor if meaningful.
 
-All commits must follow [commit-messages.mdc](commit-messages.mdc). Never use `--no-verify`.
+All commits must follow the commit message standard in `CLAUDE.md`. Never use `--no-verify`.
 
 Each phase gets its own commit so the git history proves TDD compliance.