docs(agents): short-lived worktrees and one regression test per fix

Agent-owned work now lands on main from the agent's own short-lived worktree,
cut from origin/main and removed after the push, instead of every agent sharing
one checkout. A bug fix, or a new contract between components, may add the one
test that would have caught it; there is still no TDD and no coverage for its
own sake. AGENTS.md and the cw-land, cw-orient and cw-slice skills say the
same thing.

Docs only; no build or test run.

Co-Authored-By: Claude Opus 5.5 <[email protected]>

AGENTS.md

@@ -55,10 +55,13 @@ base prompt". Two more corollaries earned here:
 - Search for behavior and symbols before reviving work from an old branch. If a
   lane is obsolete, preserve its intent and evidence rather than merging stale
   code mechanically.
-- Agent-owned work uses the canonical checkout on current `main`. Do not
-  create branches, worktrees, nested clones or task-specific integration lanes
-  to handle dirty, stale or conflicting work. When agents share the checkout,
-  partition by file and retry an `index.lock` failure. Verified main slices may
+- Agent-owned work lands on current `main`. Each writing agent works in its
+  own short-lived git worktree cut from `origin/main`, rebases and pushes
+  `main` the same day, then removes the worktree and its build directory. No
+  long-lived branches, nested clones or integration lanes, and never a new
+  worktree to hide dirty, stale or conflicting work: that work is still yours
+  to finish. In a shared checkout, stage by explicit path and retry an
+  `index.lock` failure. Verified main slices may
   be committed and pushed normally under the rule below; tag, release,
   deployment and destructive cleanup still require their explicit
   authorization.
@@ -201,10 +204,10 @@ the work even when credit is preserved.
 - Product intent and observed runtime behavior outrank a test's preferred
   implementation shape. Fix the product; do not contort production code to
   preserve a brittle assertion.
-- **No new tests and no new code comments unless explicitly asked**
-  (founder, 2026-10-08). Do not write new unit or integration tests unless the
-  user explicitly instructs it for that task; prove a change by running it and
-  the existing checks. Never write tests first and never practice TDD — this
+- **A regression test per fix; no new code comments.** Prove a change by
+  running it and the existing checks. A bug fix, or a new contract between
+  components, may add the one test that would have caught it; no coverage for
+  its own sake. Never write tests first and never practice TDD — this
   overrides any skill or default that mandates it, including superpowers
   `test-driven-development`. Existing tests stay the gate before a push: when
   a change makes one wrong, update or delete it in the same slice rather than

docs/skills/cw-land/SKILL.md

@@ -24,8 +24,9 @@ Stage 5 of the loop: [cw-orient](../cw-orient/SKILL.md) →
 
 ## Workflow
 
-1. **Land on the canonical checkout's `main`.** `AGENTS.md` owns this: no
-   maintainer PR branches, worktrees, nested clones or integration lanes.
+1. **Land on `main`.** `AGENTS.md` owns this: work in your own short-lived
+   worktree, rebase onto `origin/main`, push `HEAD:main`, then remove the
+   worktree. No maintainer PR branches, nested clones or integration lanes.
    Several agents share the checkout, so partition by file, check
    `git status` before staging, stage only your paths (never `git add -A`),
    and retry a commit that fails on `index.lock`. Push verified `main`

docs/skills/cw-orient/SKILL.md

@@ -36,7 +36,8 @@ This is stage 1 of the loop: **orient → [cw-slice](../cw-slice/SKILL.md) →
 
 2. **Look for stranded work, not lanes.** `git worktree list` and recent
    branches tell you what may hold unlanded commits. Inventory them; do not
-   start work in one, and do not create one.
+   start work in someone else's. Create your own short-lived worktree from
+   `origin/main` only after the inventory.
    ```bash
    git worktree list
    git branch --sort=-committerdate --format='%(committerdate:short) %(refname:short)' | head -20
@@ -83,7 +84,7 @@ This is stage 1 of the loop: **orient → [cw-slice](../cw-slice/SKILL.md) →
   describes reality.
 - Don't clean, stash, reset, or `git checkout --` dirt in the checkout; read
   it, prove it, and commit it, or hand it to the live writer.
-- Don't create a branch, worktree or clone to escape dirt or a conflict.
+- Don't use a new worktree to escape dirt or a conflict; inherited work is still yours.
 - Don't skip step 4 because the root `AGENTS.md` "probably covers it". The
   scoped files carry the rules that actually get violated.
 - Don't fetch, browse, or hit GitHub when the task was scoped local-only.

docs/skills/cw-slice/SKILL.md

@@ -83,8 +83,9 @@ Stage 2 of the loop: [cw-orient](../cw-orient/SKILL.md) → **slice** →
    - whether it is cross-cutting enough to need the full sweep in
      [cw-gates](../cw-gates/SKILL.md).
 
-7. **Write the implementation, then prove it runs.** No new tests or code
-   comments unless explicitly asked (`AGENTS.md`); update or delete existing
+7. **Write the implementation, then prove it runs.** A bug fix may add the
+   one regression test that would have caught it; no other new tests and no
+   new code comments (`AGENTS.md`); update or delete existing
    tests the change makes wrong.
 
 ## Red flags / don't
@@ -95,7 +96,8 @@ Stage 2 of the loop: [cw-orient](../cw-orient/SKILL.md) → **slice** →
   lifecycle system) without moving its guard test and consumers with it.
   The repo has one of each on purpose; a silent second one is the failure
   mode, not the refactor.
-- Don't write new tests unless explicitly asked, and never first. Don't add
+- Don't write tests first, and don't add tests beyond the one regression
+  test a fix earns. Don't add
   new code comments.
 - Don't contort production code to keep a brittle assertion green. A test that
   only encodes old behavior is evidence, not a veto: change it with the code.