acstack 0.49.1 → 0.50.0

package/skills/setup/SKILL.md

@@ -166,11 +166,11 @@ Add three options at the end of the list, whatever the check reported:
     So this option is the only way one of them reaches a question.
 
 - **Change one setting** – for a key the check has no opinion about, such as
-  `workflow.tdd.strict`. Point the user at `docs/config-reference.md`, which
+  `workflow.maps.persist`. Point the user at `docs/config-reference.md`, which
   lists every key, its values and its default. Then write the key:
 
   ```bash
-  ~/.claude/acstack/bin/acstack config set workflow.tdd.strict false
+  ~/.claude/acstack/bin/acstack config set workflow.maps.persist true
   ```
 
   Do not restate the reference here.

package/skills/setup/first-run.md

@@ -18,7 +18,7 @@ Profile:      [detected]
 Directories:  src/, tests/
 Commands:     [from profile]
 Frontend:     [if applicable]
-Workflow:     strict TDD, strict orchestrator, spec sync on, grilling via question tool
+Workflow:     adaptive verification, spec sync on, grilling via question tool
 ```
 
 ## Step 4: Ask what to customise
@@ -46,8 +46,6 @@ numbered list. It holds two questions in one call:
       "multiSelect": true,
       "options": [
         {"label": "Grilling", "description": "How grilling asks its questions – a question tool (default) or plain prose"},
-        {"label": "TDD strictness", "description": "Relaxed writes the tests after the code; all three phases still run"},
-        {"label": "Orchestrator", "description": "Allow direct code work instead of subagent delegation"},
         {"label": "Spec sync", "description": "Disable automatic spec sync on commit"}
       ]
     }
@@ -71,17 +69,10 @@ user did not select.
 - **Grilling** – ask `tool`, the host's question tool, against `inline` prose.
   Recommend `tool`. It writes `workflow.grilling.questions`.
 
-- **TDD strictness** – ask strict against relaxed, and write
-  `workflow.tdd.strict`.
+- Apply the shared Implementation discipline in `orchestrator/AGENTS.md` without adding workflow switches
+- Do not offer or generate the deprecated TDD booleans
 
-    Strict runs the test-writer first, so the tests exist before the code.
-    Relaxed runs the task-executor first and the test-writer after it. Relaxed
-    drops no phase, and all three agents run under both.
 
-    Recommend strict, because the relaxed test-writer works beside an
-    implementation it must not read.
-
-- **Orchestrator** – ask strict against relaxed.
 - **Spec sync** – ask enabled against disabled.
 - **Python venv** – ask the activation command. Ask it only for a Python profile.
 

package/skills/setup/report.md

@@ -13,8 +13,7 @@ Profile:      [profile]
 Source:       [source dir]
 Tests:        [tests dir]
 Grilling:     [question tool/inline prose]
-TDD:          [strict/relaxed]
-Orchestrator: [strict/relaxed]
+Verification: adaptive (shared Implementation discipline)
 Spec sync:    [enabled/disabled]
 Frontend:     [configured/not configured]
 Browser:      [project's preferred mode / not configured]
@@ -30,14 +29,10 @@ Health check: [n] pass, [n] fail, [n] info
 Next: describe what you want to build, or run /acstack:bootstrap for existing code.
 ```
 
-Fill the `Host:` line and the `Orchestrator:` line from two different sources.
-`Orchestrator:` is this project's delegation policy, read from
-`workflow.tdd.orchestrator_strict` in `acstack/project.yaml`. It says whether
-the coordinator may write code itself. `Host:` is about the machine. It names
-which tool owns terminal panes and git worktrees, and which browser modes work
-here.
-
-The `host` skill declares both, and the global config holds them.
+- Fill `Host:` from the machine declarations in global config
+- The `host` skill owns terminal panes, worktrees, and available browser modes
+- Existing deprecated workflow booleans do not control verification or delegation, including explicit `true`
+- Do not rewrite existing configuration merely because it contains those keys
 
 Fill `Workspace:` from `host.terminal_workspace`: `yes` on `true`, `declined` on
 `false`, and `not asked` when it is unset.

package/skills/setup/row-fixes.md

@@ -50,13 +50,10 @@ frontend and has not turned browser verification on. Turn it on:
 read it. Without it, every path through spec sync skips in silence, and the
 living specs never update.
 
-Launch a **read-only codebase-search subagent** over `directories.source` to propose domains.
-Use `Explore` on Claude Code and `scout` on pi.
+Investigate `directories.source` directly or through useful delegation under the shared Implementation discipline in `orchestrator/AGENTS.md`.
+Propose top-level functional areas with a kebab-case name, their paths, and one line saying what each owns.
 
-Ask it for the top-level functional areas.
-Ask for a kebab-case name, the paths that belong to it, and one line saying what it owns.
-
-Guidance for the subagent:
+Guidance for the proposal:
 
 - Group by responsibility, not by layer. Use `auth`, `billing` and `search`,
   never `models`, `controllers` and `utils`.

package/skills/implement/relaxed-tdd.md

@@ -1,28 +0,0 @@
-# Relaxed TDD
-
-This file holds what changes in step 5 of `SKILL.md` when
-`acstack/project.yaml` sets `workflow.tdd.strict: false`. Read it only then.
-
-`workflow.tdd.strict: false`, relaxed TDD, runs step 2 before step 1. It
-skips nothing. Relaxed TDD means nobody writes the tests first, and the
-first three subagents still run.
-
-The frontend-verifier's gate is the same under either setting.
-
-The format step stays immediately before the checker, so under relaxed
-it formats the new tests as well.
-
-Relaxed costs the test-writer some of its independence, and you state that
-cost rather than hide it. The implementation already sits in the working
-tree when the test-writer runs. Tell it to write the tests from the node's
-requirements, and to leave the implementation unread.
-
-Under strict there is no implementation for it to read, which is the
-stronger guarantee.
-
-The checker's gate moves with the setting. Under `true` the new tests failed
-before the executor and pass after it.
-
-Under `false` nothing can show them failing without the implementation. The
-checker then holds the new tests to existing and passing. It holds the full
-suite to passing, and the test count to not dropping.