claude-dev-env 2.4.0 → 2.5.0

package/skills/pr-converge/reference/examples.md

@@ -18,7 +18,7 @@ resets push-invalidated markers per [ground-rules.md](ground-rules.md) /
 at 270s, re-runs the sweep next tick]
 </example>
 
-<example> CODE_REVIEW tick, static sweep clean, `/code-review xhigh --fix`
+<example> CODE_REVIEW tick, static sweep clean, `/code-review ultra --fix`
 applies fixes to the working tree. Claude: [commits the applied fixes in one
 commit, pushes, resets push-invalidated markers per
 [ground-rules.md](ground-rules.md) / [state-schema.md](state-schema.md) (all
@@ -26,7 +26,7 @@ commit, pushes, resets push-invalidated markers per
 stays `phase = CODE_REVIEW`, Step 4 at 270s, returns]
 </example>
 
-<example> CODE_REVIEW tick, static sweep clean and `/code-review xhigh --fix`
+<example> CODE_REVIEW tick, static sweep clean and `/code-review ultra --fix`
 clean (no changes applied). Claude: [sets `code_review_clean_at = HEAD`,
 `phase = BUGTEAM`, runs `Skill({skill: "bugteam", ...})` in same tick]
 </example>
@@ -94,7 +94,7 @@ applies **Convergence** from `workflows/schedule-wakeup-loop.md`]
 
 <example> CODE_REVIEW tick, review body says "found 3 potential issues"
 against HEAD (a stale prior finding) but the diff is clean. Claude: [the static
-sweep and `/code-review xhigh --fix` both pass, sets `code_review_clean_at =
+sweep and `/code-review ultra --fix` both pass, sets `code_review_clean_at =
 HEAD`, `phase = BUGTEAM`]
 </example>
 

package/skills/pr-converge/reference/fix-protocol.md

@@ -32,6 +32,6 @@ files during fix phase in multi-PR mode.
 regardless of phase. A new commit **resets the full convergence cycle**: a
 code-review clean, a bugteam clean, and a Bugbot clean on an older SHA do
 **not** count toward convergence on the new `HEAD`. Re-run the static sweep and
-`/code-review xhigh --fix` on `current_head`, then bugteam, then the terminal
+`/code-review ultra --fix` on `current_head`, then bugteam, then the terminal
 Bugbot gate, all on the same `HEAD` with no intervening push. Re-entering in the
 same tick saves a wakeup cycle.

package/skills/pr-converge/reference/ground-rules.md

@@ -10,11 +10,14 @@
 - **All `*_clean_at`, `merge_state_status`, and `bugbot_down` / `codex_down` reset on every push.**
 - **`bugbot run` comment is load-bearing.** Literal phrase exactly —
   empirically the only re-trigger Cursor Bugbot recognizes.
-- **Production edits go through `clean-coder`, except `/code-review xhigh --fix`.**
+- **Production edits go through `clean-coder`, except `/code-review ultra --fix`.**
   The lead never hand-edits production files. Every bugbot, bugteam,
-  Copilot, or Claude finding spawns `Agent(subagent_type="clean-coder")` to
+  Copilot, or Claude finding spawns `Agent(subagent_type="clean-coder",
+  model="sonnet")` — worker-model routing per
+  [`skills/orchestrator/SKILL.md`](../../orchestrator/SKILL.md#workflow-agent-routing);
+  resolver-supplied sonnet-equivalent on third-party hosts — to
   apply the fix. The CODE_REVIEW phase is the one exception: `/code-review
-  xhigh --fix` applies its own findings to the working tree, which the next
+  ultra --fix` applies its own findings to the working tree, which the next
   CODE_REVIEW/BUGTEAM cycle re-reviews after the loop resets.
 - **Adapt when reality contradicts on-disk state.** If `state.json`,
   `git`, or `gh` disagree with live PR, escalate as hard blocker per
@@ -22,7 +25,7 @@
 - **Cross-repo cwd routing is routine, not a fork.** When the PR under
   convergence lives in a different repo than the session is rooted in, route
   the working directory into a checkout of the PR's repo automatically —
-  `/code-review xhigh --fix`, `git`, and every `clean-coder` fix spawn act on the
+  `/code-review ultra --fix`, `git`, and every `clean-coder` fix spawn act on the
   repo of the current working directory. The resolution is fixed
   ([per-tick.md § Step 1.5](per-tick.md)): resolve the PR worktree, `cd` into
   it, run local work there. Do not pause, ask, or raise it as a material fork.

package/skills/pr-converge/reference/multi-pr-orchestration.md

@@ -108,7 +108,10 @@ one agent per PR in single parallel message. Never process any PR inline.
 Bugfind subagent completes (findings or clean):
 
 - **PRs with findings:** spawn one fix worker per PR via
-  `Agent(subagent_type="clean-coder", run_in_background=true)`. Worker:
+  `Agent(subagent_type="clean-coder", run_in_background=true,
+  model="sonnet")` — worker-model routing per
+  [`skills/orchestrator/SKILL.md`](../../orchestrator/SKILL.md#workflow-agent-routing);
+  resolver-supplied sonnet-equivalent on third-party hosts. Worker:
   1. Reads outcomes XML.
   2. Applies TDD fixes (test first, then production).
   3. Commits, pushes one fix commit.

package/skills/pr-converge/reference/per-tick.md

@@ -71,7 +71,7 @@ Capture `number`, `head.sha` (= `current_head`), owner/repo, branch.
 
 The **PR worktree** is the local working tree of the PR's repo on its head
 branch. Every local operation this tick runs there: the CODE_REVIEW static sweep
-and `/code-review xhigh --fix`, every `clean-coder` fix spawn, and every commit and
+and `/code-review ultra --fix`, every `clean-coder` fix spawn, and every commit and
 push. `/code-review` and `git` both act on the repo of the current working
 directory, so the working directory must be the PR worktree before any local
 work begins. Re-resolve it every tick — a rebase or a fresh HEAD can move the
@@ -180,7 +180,7 @@ CODE_REVIEW.
 ### `phase == CODE_REVIEW`
 
 The entry phase of every convergence tick, re-entered after any fix push. It runs
-a deterministic static sweep, then the built-in `/code-review xhigh --fix` on the
+a deterministic static sweep, then the built-in `/code-review ultra --fix` on the
 full `origin/main...HEAD` diff at effort high on model opus. `/code-review`
 produces no GitHub review artifact, so there are no code-review threads to
 resolve.
@@ -196,7 +196,7 @@ a. **Static sweep — runs first, before `/code-review`.** Run the deterministic
    `phase = CODE_REVIEW`, and re-run the sweep. When the sweep is clean, run
    the host-aware review below.
 
-b. Run the built-in `/code-review xhigh --fix` with OPUS on the FULL `origin/main...HEAD`
+b. Run the built-in `/code-review ultra --fix` with OPUS on the FULL `origin/main...HEAD`
    diff — every file the PR touches — via the
    [local diff review](https://code.claude.com/docs/en/code-review#review-a-diff-locally).
    The review always runs at effort high on model opus. It reviews the diff and
@@ -231,14 +231,14 @@ b. Run the built-in `/code-review xhigh --fix` with OPUS on the FULL `origin/mai
    Match the first mode whose predicate holds:
 
    - **`mode == "in_session"`** (Claude host and session model is opus): run
-     `/code-review xhigh --fix` with OPUS in this session with no path arguments so it
+     `/code-review ultra --fix` with OPUS in this session with no path arguments so it
      audits the whole branch diff against `origin/main`. After it returns, a
      non-empty `git status --porcelain` means fixes applied (`dirty_tree`
      equivalent). Treat a failed in-session slash command the same as a failed
      review: do not set `code_review_clean_at`.
    - **`mode == "chain"`** (any other host, or a Claude session on any model
      other than opus): the helper already ran the headless review
-     (`claude -p "/code-review xhigh --fix" --model opus` through the chain
+     (`claude -p "/code-review ultra --fix" --model opus` through the chain
      runner) with cwd set to the PR worktree. Read `returncode`,
      `served_command`, and `dirty_tree` from the JSON. A successful serve is
      `returncode == 0` with a non-null `served_command`. `dirty_tree` true

package/skills/pr-converge/reference/progress-checklist.md

@@ -91,7 +91,7 @@ round as converged. This rule holds every tick, every loop, every PR.
       - [ ] **Static sweep fails** → apply shared fix protocol → push → reset markers
             → stay CODE_REVIEW → Step 5
       - [ ] **`mode == in_session`** (Claude host, session model opus) → run
-            `/code-review xhigh --fix` in-session (no path args)
+            `/code-review ultra --fix` in-session (no path args)
       - [ ] **`mode == chain`** (any other host or non-opus session) → helper
             already ran the headless review; read `returncode`,
             `served_command`, and `dirty_tree` from JSON