feat: sync generic context capabilities

.agents/skills/plan-go/scripts/loop_spec.sh

@@ -26,9 +26,9 @@ $goal
 - Confirm the plan is Aligned and execution is allowed.
 - Treat that registered aligned plan as the goal source and implementation facts as owned by the target repository.
 - If the gate fails, route to task-plan or stop for alignment before substantive work.
-- Record goal tooling as an object with a non-empty runtime, `used=true`, and `mechanism=task-board-plan` when the registered task/plan is the goal.
-- Use `mechanism=native-goal+task-board-plan` only when a native runtime goal mirrors that registered task/plan.
-- Keep `task_id` as the auditable registered goal identifier; the runtime name does not imply native goal usage.
+- Record goal tooling as an object with a non-empty runtime, 'used=true', and 'mechanism=task-board-plan' when the registered task/plan is the goal.
+- Use 'mechanism=native-goal+task-board-plan' only when a native runtime goal mirrors that registered task/plan.
+- Keep 'task_id' as the auditable registered goal identifier; the runtime name does not imply native goal usage.
 - Reject chat-only, verbal, or arbitrary goal-mechanism claims.
 
 ## Acceptance Checks
@@ -42,7 +42,7 @@ $goal
 ## Role Independence Policy
 - Main orchestrates and integrates: confirm the goal, hand off prompts, integrate accepted output, run mechanical checks, record evidence, and choose routes.
 - Executor produces the smallest sufficient change and reports artifacts, commands/results, validation entry points, and uncertainties.
-- Executor artifacts must identify either a safe existing repository-relative `local_path` or an HTTP(S) `external_url`; bare strings are not evidence.
+- Executor artifacts must identify either a safe existing repository-relative 'local_path' or an HTTP(S) 'external_url'; bare strings are not evidence.
 - Evaluator pressure-tests the result against the original goal and reports categorized findings with evidence.
 - Main must not claim executor or evaluator output as its own.
 - If independent agent tools are available, use separate executor and evaluator agents. Main-thread implementation plus self-review is not a complete adversarial loop.

.agents/skills/plan-go/tests/test_plan_go.py

@@ -209,9 +209,16 @@ def test_adapted_source_retains_mit_notice(self):
     def test_loop_spec_is_repository_relative_and_names_roles(self):
         result = run(SPEC, "Execute the plan")
         self.assertEqual(0, result.returncode, result.stderr)
+        self.assertEqual("", result.stderr)
         self.assertIn("Execute the plan", result.stdout)
         self.assertIn("Executor Prompt", result.stdout)
         self.assertIn("Evaluator Prompt", result.stdout)
+        self.assertIn("'used=true'", result.stdout)
+        self.assertIn("'mechanism=task-board-plan'", result.stdout)
+        self.assertIn("'mechanism=native-goal+task-board-plan'", result.stdout)
+        self.assertIn("'task_id'", result.stdout)
+        self.assertIn("'local_path'", result.stdout)
+        self.assertIn("'external_url'", result.stdout)
         self.assertIn("out-of-scope findings as follow-up work", result.stdout.lower())
         self.assertIn("acceptance conditions are satisfied", result.stdout.lower())
         self.assertNotIn("~/.codex", result.stdout)

.agents/skills/task-board/SKILL.md

@@ -28,10 +28,12 @@ scripts/task validate
 scripts/task list [--status STATUS] [--format table|yaml|json] [--tree]
 scripts/task show TASK_ID [--format yaml|json]
 scripts/task add-task --id ID --title TITLE --domain DOMAIN [options]
+scripts/task set-title TASK_ID TITLE
 scripts/task register-plan PLAN --id ID --domain DOMAIN --repo REPO [options]
 scripts/task set-status TASK_ID STATUS
 scripts/task set-plan-metadata TASK_ID PLAN [routing options]
 scripts/task add-note TASK_ID NOTE
+scripts/task remove-note TASK_ID NOTE
 scripts/task add-evidence TASK_ID --type TYPE --status STATUS --summary TEXT [--path PATH]
 scripts/task link-sub-task PARENT_TASK_ID CHILD_TASK_ID
 scripts/task unlink-sub-task PARENT_TASK_ID CHILD_TASK_ID

.agents/skills/task-board/scripts/task

@@ -665,6 +665,29 @@ def cmd_set_status(args: argparse.Namespace) -> int:
     return 0
 
 
+def cmd_set_title(args: argparse.Namespace) -> int:
+    board, yaml, tasks, _statuses = checked_board(args.board)
+    task = find_task(tasks, args.task_id)
+    title = args.title.strip()
+    if not title:
+        raise TaskError("title must be non-empty")
+    old_title = task.get("title")
+    if old_title == title:
+        print(f"unchanged: {task['id']} title already matches")
+        return 0
+    now = utc_now()
+    task["title"] = title
+    task["updated_at"] = now
+    board["updated_at"] = now
+    _tasks, _statuses, errors = validate_board(board)
+    if errors:
+        task["title"] = old_title
+        raise TaskError("title change would make board invalid:\n" + "\n".join(f"- {error}" for error in errors))
+    write_board(args.board, board, yaml)
+    print(f"updated: {task['id']} title")
+    return 0
+
+
 def cmd_set_plan_metadata(args: argparse.Namespace) -> int:
     board, yaml, tasks, statuses = checked_board(args.board)
     task = find_task(tasks, args.task_id)
@@ -874,6 +897,28 @@ def cmd_add_note(args: argparse.Namespace) -> int:
     return 0
 
 
+def cmd_remove_note(args: argparse.Namespace) -> int:
+    board, yaml, tasks, _statuses = checked_board(args.board)
+    task = find_task(tasks, args.task_id)
+    notes = task.setdefault("notes", [])
+    if args.note not in notes:
+        print(f"unchanged: {task['id']} does not have this note")
+        return 0
+    notes.remove(args.note)
+    now = utc_now()
+    task["updated_at"] = now
+    board["updated_at"] = now
+    _tasks, _statuses, errors = validate_board(board)
+    if errors:
+        raise TaskError(
+            "note removal would make board invalid:\n"
+            + "\n".join(f"- {error}" for error in errors)
+        )
+    write_board(args.board, board, yaml)
+    print(f"updated: {task['id']} notes -= 1")
+    return 0
+
+
 def cmd_link_sub_task(args: argparse.Namespace) -> int:
     board, yaml, tasks, _statuses = checked_board(args.board)
     parent = find_task(tasks, args.parent_task_id)
@@ -1069,6 +1114,12 @@ def build_parser() -> argparse.ArgumentParser:
     set_status.add_argument("status")
     set_status.set_defaults(func=cmd_set_status)
 
+    set_title = subparsers.add_parser("set-title", help="Rename a task title.")
+    add_board_arg(set_title)
+    set_title.add_argument("task_id")
+    set_title.add_argument("title")
+    set_title.set_defaults(func=cmd_set_title)
+
     set_plan_metadata = subparsers.add_parser(
         "set-plan-metadata",
         help="Refresh an existing plan link/status and atomically update its execution routing.",
@@ -1149,6 +1200,12 @@ def build_parser() -> argparse.ArgumentParser:
     add_note.add_argument("note")
     add_note.set_defaults(func=cmd_add_note)
 
+    remove_note = subparsers.add_parser("remove-note", help="Remove an exact note from a task.")
+    add_board_arg(remove_note)
+    remove_note.add_argument("task_id")
+    remove_note.add_argument("note")
+    remove_note.set_defaults(func=cmd_remove_note)
+
     link_sub_task = subparsers.add_parser(
         "link-sub-task", help="Link an existing task as a child of another task."
     )

.agents/skills/task-board/tests/test_task_cli.py

@@ -202,7 +202,19 @@ def test_add_note_evidence_and_status_transitions_are_idempotent(self) -> None:
                 "add-note", "example.release", "Synthetic review note."
             )
             self.assertEqual(result, 0, stderr)
-        self.assertEqual(self.task("example.release")["notes"], ["Synthetic review note."])
+            self.assertEqual(self.task("example.release")["notes"], ["Synthetic review note."])
+
+        for _ in range(2):
+            result, _stdout, stderr = self.run_cli(
+                "remove-note", "example.release", "Synthetic review note."
+            )
+            self.assertEqual(result, 0, stderr)
+        self.assertEqual(self.task("example.release")["notes"], [])
+
+        result, _stdout, stderr = self.run_cli(
+            "add-note", "example.release", "Synthetic review note."
+        )
+        self.assertEqual(result, 0, stderr)
 
         evidence_args = (
             "add-evidence",
@@ -230,6 +242,25 @@ def test_add_note_evidence_and_status_transitions_are_idempotent(self) -> None:
         self.assertEqual(result, 0, stderr)
         self.assertEqual(self.task("example.release")["status"], "done")
 
+    def test_set_title_is_validated_and_idempotent(self) -> None:
+        self.add_task("example.rename")
+        result, _stdout, stderr = self.run_cli(
+            "set-title", "example.rename", "Renamed task"
+        )
+        self.assertEqual(result, 0, stderr)
+        self.assertEqual(self.task("example.rename")["title"], "Renamed task")
+
+        result, stdout, stderr = self.run_cli(
+            "set-title", "example.rename", "Renamed task"
+        )
+        self.assertEqual(result, 0, stderr)
+        self.assertIn("unchanged", stdout)
+
+        result, _stdout, stderr = self.run_cli("set-title", "example.rename", "   ")
+        self.assertEqual(result, 1)
+        self.assertIn("non-empty", stderr)
+        self.assertEqual(self.task("example.rename")["title"], "Renamed task")
+
     def test_status_failure_and_blocked_intake_are_atomic(self) -> None:
         self.add_task("example.unplanned")
         before = self.board_path.read_bytes()
@@ -252,6 +283,20 @@ def test_status_failure_and_blocked_intake_are_atomic(self) -> None:
         self.assertEqual(result, 0, stderr)
         self.assertTrue(self.task("example.blocked")["notes"])
 
+    def test_remove_sole_blocker_note_is_rejected_without_writes(self) -> None:
+        self.add_task("example.blocked-note", status="blocked")
+        notes_before = list(self.task("example.blocked-note")["notes"])
+        self.assertEqual(len(notes_before), 1)
+        before = self.board_path.read_bytes()
+
+        result, _stdout, stderr = self.run_cli(
+            "remove-note", "example.blocked-note", notes_before[0]
+        )
+        self.assertEqual(result, 1)
+        self.assertIn("would make board invalid", stderr)
+        self.assertEqual(self.board_path.read_bytes(), before)
+        self.assertEqual(self.task("example.blocked-note")["notes"], notes_before)
+
     def test_register_plan_creates_and_attaches_with_strict_pairing(self) -> None:
         aligned = self.make_plan("register.md", title="Checkout registration")
         result, _stdout, stderr = self.run_cli(

.gitignore

@@ -13,6 +13,7 @@ tmp/
 *.swp
 
 # Local-only material must not be published.
+.playwright-mcp/
 .env
 .env.*
 !.env.example

init.sh

@@ -32,6 +32,9 @@ files=(
   scripts/task
   tasks/board.yaml
   tasks/task.schema.json
+  tests/__init__.py
+  tests/helpers.py
+  tests/test_claude_compat.py
   .agents/skills/task-board/SKILL.md
   .agents/skills/task-board/scripts/task
   .agents/skills/task-plan/SKILL.md

tests/test_init.py

@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import os
+import sys
 import tempfile
 import unittest
 from pathlib import Path
@@ -26,6 +27,9 @@ def test_initializes_an_empty_workspace_with_spaces(self) -> None:
                 "scripts/task",
                 "tasks/board.yaml",
                 "tasks/task.schema.json",
+                "tests/__init__.py",
+                "tests/helpers.py",
+                "tests/test_claude_compat.py",
                 ".agents/skills/task-board/SKILL.md",
                 ".agents/skills/task-plan/SKILL.md",
                 ".agents/skills/plan-go/SKILL.md",
@@ -65,6 +69,16 @@ def test_initializes_an_empty_workspace_with_spaces(self) -> None:
             validation = run(destination / "scripts/task", "validate", cwd=destination)
             self.assertEqual(0, validation.returncode, validation.stderr)
 
+            compatibility = run(
+                sys.executable,
+                "-m",
+                "unittest",
+                "tests.test_claude_compat",
+                "-v",
+                cwd=destination,
+            )
+            self.assertEqual(0, compatibility.returncode, compatibility.stderr)
+
     def test_repeated_initialization_is_idempotent(self) -> None:
         with tempfile.TemporaryDirectory() as directory:
             destination = Path(directory) / "project"