diff --git a/.claude/settings/permissions/bash.jsonc b/.claude/settings/permissions/bash.jsonc
index 92caa8a2..bdf27717 100644
--- a/.claude/settings/permissions/bash.jsonc
+++ b/.claude/settings/permissions/bash.jsonc
@@ -74,9 +74,17 @@
       "Bash(tail *)",
       // Search
       "Bash(rg *)",
+      // Research
+      "Bash(gh issue list *)",
+      "Bash(gh pr view *)",
+      "Bash(gh pr diff *)"
     ],
     "ask": [
-      "Bash(gh *)", // let's hold off before we let it use the github CLI in any free running allow mode...I don't want it somehow approving PRs with the user's credentials
+      // let's hold off before we let it use the github CLI in any free running allow mode...I don't want it somehow approving PRs with the user's credentials
+      "Bash(gh repo *)",
+      "Bash(gh release *)",
+      "Bash(gh secret *)",
+      "Bash(gh ruleset *)",
       "Bash(aws *)", // let's hold off before we let it use AWS CLI in any free running allow mode. We need to be very sure we don't have any access to staging or production credentials in our dev environment (...which we shouldn't...but we need to double check that or consider any other safeguards first)
       "Bash(curl *)",
       "Bash(ln *)",
@@ -85,6 +93,17 @@
     "deny": [
       // Exceptions to generally allowed AI tooling
       "Bash(bd init*)", // we need to control the init process, don't let AI do that in the background
+      // Github
+      // Claude should not ever interfere with the PR process, that is how we gate AI's work
+      "Bash(gh pr create *)",
+      "Bash(gh pr edit *)",
+      "Bash(gh pr ready *)",
+      "Bash(gh pr review *)",
+      "Bash(gh pr merge *)",
+      "Bash(gh pr close *)",
+      "Bash(gh pr comment *)",
+      "Bash(gh pr update-branch *)",
+
       // Destructive File Operations
       "Bash(chmod -R *)",
       "Bash(chown -R *)",
diff --git a/.devcontainer/devcontainer.json b/.devcontainer/devcontainer.json
index 4922e7cb..78f871bc 100644
--- a/.devcontainer/devcontainer.json
+++ b/.devcontainer/devcontainer.json
@@ -75,5 +75,5 @@
   "initializeCommand": "sh .devcontainer/initialize-command.sh",
   "onCreateCommand": "sh .devcontainer/on-create-command.sh",
   "postStartCommand": "sh .devcontainer/post-start-command.sh"
-  // Devcontainer context hash (do not manually edit this, it's managed by a pre-commit hook): 01a288c4 # spellchecker:disable-line
+  // Devcontainer context hash (do not manually edit this, it's managed by a pre-commit hook): 5ecc43b1 # spellchecker:disable-line
 }
diff --git a/.devcontainer/manual-setup-deps.py b/.devcontainer/manual-setup-deps.py
index 6f6fe0d8..53e59e1b 100644
--- a/.devcontainer/manual-setup-deps.py
+++ b/.devcontainer/manual-setup-deps.py
@@ -44,6 +44,12 @@
     default=False,
     help="Allow uv to install new versions of Python on the fly. This is typically only needed when instantiating the copier template.",
 )
+_ = parser.add_argument(
+    "--skip-installing-pulumi-cli",
+    action="store_true",
+    default=False,
+    help="Do not install the Pulumi CLI even if the lock file references it",
+)
 
 
 class PackageManager(str, enum.Enum):
@@ -127,6 +133,17 @@ def main():
                     check=True,
                     env=uv_env,
                 )
+            if (
+                not generate_lock_file_only
+                and not args.skip_installing_pulumi_cli
+                and platform.system() == "Linux"
+                and env.lock_file.exists()
+                and '"pulumi"' in env.lock_file.read_text()
+            ):
+                _ = subprocess.run(
+                    ["sh", str(REPO_ROOT_DIR / ".devcontainer" / "install-pulumi-cli.sh"), str(env.lock_file)],
+                    check=True,
+                )
         elif env.package_manager == PackageManager.PNPM:
             pnpm_command = ["pnpm", "install", "--dir", str(env.path)]
             if env_check_lock:
diff --git a/AGENTS.md b/AGENTS.md
index 115db424..6c9e9f64 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -17,7 +17,8 @@ This project is a Copier template used to generate other copier templates. It is
 
 ## Testing
 
-- Always run tests with an explicit path (e.g. uv run pytest tests/unit) — test runners discover all types by default.
+- Always run tests with an explicit path (e.g. uv run pytest tests/unit) — test runners discover all types (unit, integration, E2E...) by default.
+- When iterating on a single test, run that test in isolation first and confirm it is in the expected state (red or green) before widening to the full suite. Use the most targeted invocation available: a specific test function for Python (e.g. `uv run pytest path/to/test.py::test_name --no-cov`) or a file path and name filter for TypeScript (e.g. `pnpm test-unit -- path/to/test.spec.ts -t "test name" --no-coverage`). Only run the full suite once the target test behaves as expected.
 - Test coverage requirements are usually at 100%, so when running a subset of tests, always disable test coverage to avoid the test run failing for insufficient coverage.
 - Avoid magic values in comparisons in tests in all languages (like ruff rule PLR2004 specifies)
 - Prefer using random values in tests rather than arbitrary ones (e.g. the faker library, uuids, random.randint) when possible. For enums, pick randomly rather than hardcoding one value.
@@ -26,11 +27,12 @@ This project is a Copier template used to generate other copier templates. It is
 
 ### Python Testing
 
-- When using `mocker.spy` on a class-level method (including inherited ones), the spy records the unbound call, so assertions need `ANY` as the first argument to match self:  `spy.assert_called_once_with(ANY, expected_arg)`
+- When using `mocker.spy` on a class-level method (including inherited ones), the spy records the unbound call, so assertions need `ANY` as the first argument to match self: `spy.assert_called_once_with(ANY, expected_arg)`
 - Before writing new mock/spy helpers, check the `tests/unit/` folder for pre-built helpers in files like `fixtures.py` or `*mocks.py`
 - When a test needs a fixture only for its side effects (not its return value), use `@pytest.mark.usefixtures(fixture_name.__name__)` instead of adding an unused parameter with a noqa comment
 - Use `__name__` instead of string literals when referencing functions/methods (e.g., `mocker.patch.object(MyClass, MyClass.method.__name__)`, `pytest.mark.usefixtures(my_fixture.__name__)`). This enables IDE refactoring tools to catch renames.
 - When using the faker library, prefer the pytest fixture (provided by the faker library) over instantiating instances of Faker.
+- **Choosing between cassettes and mocks:** At the layer that directly wraps an external API or service, strongly prefer VCR cassette-recorded interactions (via pytest-recording/vcrpy) — they capture real HTTP traffic and verify the wire format, catching integration issues that mocks would miss. At layers above that (e.g. business logic, route handlers), mock the wrapper layer instead (e.g. `mocker.patch.object(ThresholdsRepository, ...)`) — there is no value in re-testing the HTTP interaction from higher up.
 - **Never hand-write VCR cassette YAML files.** Cassettes must be recorded from real HTTP interactions by running the test once with `--record-mode=once` against a live external service: `uv run pytest --record-mode=once <test path> --no-cov`. The default mode is `none` — a missing cassette will cause an error, which is expected until recorded.
 - **Never hand-edit syrupy snapshot files.** Snapshots are auto-generated — to create or update them, run `uv run pytest --snapshot-update <test path> --no-cov`. A missing snapshot causes the test to fail, which is expected until you run with `--snapshot-update`. When a snapshot mismatch occurs, fix the code if the change was unintentional; run `--snapshot-update` if it was intentional.
 - **Never hand-write or hand-edit pytest-reserial `.jsonl` recording files.** Recordings must be captured from real serial port traffic by running the test with `--record` while the device is connected: `uv run pytest --record <test path> --no-cov`. The default mode replays recordings — a missing recording causes an error, which is expected until recorded against a live device.
@@ -49,7 +51,8 @@ This project is a Copier template used to generate other copier templates. It is
 - For frontend tests, run commands via `pnpm` scripts from `frontend/package.json` — never invoke tools directly (not pnpm exec <tool>, npx <tool>, etc.). ✅ pnpm test-unit  ❌ pnpm vitest ... or npx vitest ...
 - For linting and type-checking, prefer `pre-commit run <hook-id>` over invoking tools directly — this matches the permission allow-list and mirrors what CI runs. Key hook IDs: `typescript-check`, `eslint`, `pyright`, `ruff`, `ruff-format`.
 - Never rely on IDE diagnostics for ruff warnings — the IDE may not respect the project's ruff.toml config. Run `pre-commit run ruff -a` to get accurate results.
-- When running terminal commands, execute exactly one command per tool call. Do not chain commands with &&, ||, ;, or & — this prohibition has no exceptions, even for `cd && ...` patterns. Use absolute paths instead of `cd` to avoid needing to chain. Pipes (|) are allowed for output transformation (e.g., head, tail, grep). If two sequential commands are needed, run them in separate tool calls. Chained commands break the permission allow-list matcher and cause unnecessary permission prompts
+- When running terminal commands, execute exactly one command per tool call. Do not chain commands with &&, ||, ;, or & — this prohibition has no exceptions, even for `cd && ...` patterns. Use `cd` to change to the directory you want before running the command, avoiding the need to chain. Pipes (|) are allowed for output transformation (e.g., head, tail, grep). If two sequential commands are needed, run them in separate tool calls. Chained commands break the permission allow-list matcher and cause unnecessary permission prompts
+- Never use `pnpm --prefix <path>` or `uv --directory <path>` to target a different directory — these flags break the permission allow-list matcher the same way chained `cd &&` commands do. Instead, rely on the working directory already being correct (the cwd persists between Bash tool calls), or issue a plain `cd <path>` as a separate prior tool call to reposition before running the command.
 - Never use backslash line continuations in shell commands — always write the full command on a single line. Backslashes break the permission allow-list matcher.
 - **Never manually edit files in any `generated/` folder.** These files are produced by codegen tooling (typically Kiota) and any manual changes will be overwritten. If a generated file needs to change, update the source (e.g. the OpenAPI schema) and re-run the generator.
 
diff --git a/template/.github/actions/install_deps/action.yml.jinja-base b/template/.github/actions/install_deps/action.yml.jinja-base
index 5068c331..c21e92b7 100644
--- a/template/.github/actions/install_deps/action.yml.jinja-base
+++ b/template/.github/actions/install_deps/action.yml.jinja-base
@@ -44,6 +44,11 @@ inputs:
     description: Whether to skip updating the hash when running manual-setup-deps.py
     default: true
     required: false
+  skip-installing-pulumi-cli:
+    type: boolean
+    description: Whether to skip installing the Pulumi CLI even if the lock file references it
+    default: true
+    required: false
 
 
 runs:
@@ -83,5 +88,5 @@ runs:
     - name: Install dependencies
       # the funky syntax is github action ternary
       if: ${{ inputs.install-deps }}
-      run: python .devcontainer/manual-setup-deps.py ${{ inputs.python-version == 'notUsing' && '--no-python' || '' }} ${{ inputs.node-version == 'notUsing' && '--no-node' || '' }} ${{ inputs.skip-updating-devcontainer-hash && '--skip-updating-devcontainer-hash' || '' }}
+      run: python .devcontainer/manual-setup-deps.py ${{ inputs.python-version == 'notUsing' && '--no-python' || '' }} ${{ inputs.node-version == 'notUsing' && '--no-node' || '' }} ${{ inputs.skip-updating-devcontainer-hash && '--skip-updating-devcontainer-hash' || '' }} ${{ inputs.skip-installing-pulumi-cli && '--skip-installing-pulumi-cli' || '' }}
       shell: pwsh{% endraw %}
diff --git a/template/.github/workflows/pre-commit.yaml.jinja-base b/template/.github/workflows/pre-commit.yaml.jinja-base
index 1a9c838b..7f590bb6 100644
--- a/template/.github/workflows/pre-commit.yaml.jinja-base
+++ b/template/.github/workflows/pre-commit.yaml.jinja-base
@@ -50,6 +50,7 @@ jobs:
           python-version: ${{ inputs.python-version }}
           node-version: ${{ inputs.node-version }}
           skip-installing-ssm-plugin-manager: true
+          skip-installing-pulumi-cli: true
 
       - name: Set up mutex # Github concurrency management is horrible, things get arbitrarily cancelled if queued up. So using mutex until github fixes itself. When multiple jobs are modifying cache at once, weird things can happen.  possible issue is https://github.com/actions/toolkit/issues/658
         if: ${{ runner.os != 'Windows' }} # we're just gonna have to YOLO on Windows, because this action doesn't support it yet https://github.com/ben-z/gh-action-mutex/issues/14
diff --git a/template/template/.github/workflows/{% if template_uses_pulumi %}pulumi-aws.yml{% endif %}.jinja-base b/template/template/.github/workflows/{% if template_uses_pulumi %}pulumi-aws.yml{% endif %}.jinja-base
index b924c600..eedb3f5f 100644
--- a/template/template/.github/workflows/{% if template_uses_pulumi %}pulumi-aws.yml{% endif %}.jinja-base	
+++ b/template/template/.github/workflows/{% if template_uses_pulumi %}pulumi-aws.yml{% endif %}.jinja-base	
@@ -122,6 +122,7 @@ jobs:
         uses: ./.github/actions/install_deps
         with:
           python-version: ${{ inputs.PYTHON_VERSION }}
+          skip-installing-pulumi-cli: true # at the moment, we're more specifically installing the Pulumi CLI just in the target directory inside the pulumi_ephemeral_deploy action
 
       - name: Download Artifact
         uses: actions/download-artifact@{% endraw %}{{ gha_download_artifact }}{% raw %}