plain-forge 1.0.16 → 1.0.17

package/forge/rules/examples/integration-embedded-testing/run_conformance_tests_java.sh

@@ -21,7 +21,18 @@ fi
 
 current_dir=$(pwd)
 
-CONFORMANCE_TESTS_FOLDER=".tmp/java_conformance"
+# Resolve the conformance tests folder to an absolute path ($2 may be relative
+# on older renderers; newer renderers pass it as an absolute path).
+case "$2" in
+  /*) CONFORMANCE_TESTS_SOURCE="$2" ;;
+  *)  CONFORMANCE_TESTS_SOURCE="$current_dir/$2" ;;
+esac
+
+# Scratch folder lives in the system temp directory so no build debris is left
+# inside the project. It is removed again when the script exits.
+CONFORMANCE_TESTS_FOLDER="/tmp/java_conformance"
+
+trap 'rm -rf "$CONFORMANCE_TESTS_FOLDER"' EXIT
 
 cd "$current_dir" 2>/dev/null
 
@@ -43,8 +54,8 @@ else
   mkdir -p $CONFORMANCE_TESTS_FOLDER
 fi
 
-printf "Copying all files and folders from $2 to $CONFORMANCE_TESTS_FOLDER...\n"
-cp -R $2/* $CONFORMANCE_TESTS_FOLDER
+printf "Copying all files and folders from $CONFORMANCE_TESTS_SOURCE to $CONFORMANCE_TESTS_FOLDER...\n"
+cp -R "$CONFORMANCE_TESTS_SOURCE"/* $CONFORMANCE_TESTS_FOLDER
 
 # Move to the subfolder
 printf "Moving to $CONFORMANCE_TESTS_FOLDER...\n"

package/forge/rules/integration-embedded-testing.md

@@ -10,13 +10,13 @@ When an embedded integration ships its three `test_scripts/` (prepare-environmen
 
 ## Staging model (read this first)
 
-The three scripts do **not** all stage into the same place. Embedded integrations need a runnable host project, so the prepare and unit-test scripts operate **inside the host codebase itself**; only the conformance script uses a `.tmp/` scratch folder because the conformance suite lives in a separate project.
+The three scripts do **not** all stage into the same place. Embedded integrations need a runnable host project, so the prepare and unit-test scripts operate **inside the host codebase itself**; only the conformance script uses a scratch folder in the system temp directory (`/tmp/<lang>_conformance/`) because the conformance suite lives in a separate project.
 
 | Script | Where it operates | What it does to the host source tree |
 |--------|-------------------|---------------------------------------|
 | `prepare_environment_<lang>` | Inside the host codebase | Copies `$1` into the module's package path under the host's source tree, then compiles / installs the host project |
 | `run_unittests_<lang>` | Inside the host codebase | Same copy as above (self-contained), then runs the module's unit tests + lint inside the host |
-| `run_conformance_tests_<lang>` | `.tmp/<lang>_conformance/` | Copies `$2` (the conformance-tests folder) into the scratch directory and runs the conformance suite there, which depends on the host build that `prepare_environment` already installed |
+| `run_conformance_tests_<lang>` | `/tmp/<lang>_conformance/` (system temp) | Copies `$2` (the conformance-tests folder) into the scratch directory and runs the conformance suite there, which depends on the host build that `prepare_environment` already installed |
 
 This deliberately writes into the host's `src/main/...` and `src/test/...` (or the language equivalent). Two consequences flow from that:
 
@@ -127,11 +127,13 @@ All three scripts are invoked by the `codeplain` renderer with positional argume
    - `2` — missing or inaccessible input folder
    - `69` — unrecoverable environment failure (missing toolchain, cannot enter working folder, install failed)
    - Any other non-zero code — propagated verbatim from the underlying build / test tool
+
+   The renderer passes `$1` (and `$2` for the conformance runner) as **absolute paths** (older renderer versions passed relative ones). Never build a derived path by concatenating a raw argument (`.tmp/$1`, `<lang>_$1`, `$current_dir/$2`) — with an absolute argument that produces a doubled path (`/project//abs/path/...`). Use the argument as-is where a path is needed, take its `basename` when naming a scratch folder, and resolve `$2` to absolute (if it isn't already) before any `cd`
 3. **Echo the toolchain home at the top** — the first thing a developer checks when CI breaks. For Java: `JAVA_HOME`. For Python: the active interpreter (`python --version` / `which python`). For Node: `node --version`. For Go: `go env GOROOT`. Pick the variable whose value most often explains "why does it work locally but not on the CI runner?"
 4. **Respect `VERBOSE=1`.** Gate chatty diagnostic prints behind `if [ "${VERBOSE:-}" = "1" ]` (and the PowerShell equivalent). Errors and key step markers print unconditionally
 5. **Resolve paths relative to the script, not `$PWD`.** Use `"$(cd "$(dirname "$0")/<relative-path-to-anchor>" && pwd)"`. Hard-coded `../../` chains break the moment the renderer's `cwd` changes
 6. **Create destination directories before copying.** `mkdir -p` before any `cp -R` / `rsync` / `robocopy` — most copy commands do not create intermediate directories and fail silently in some shells
-7. **Scope destructive operations narrowly.** Any `rm -rf` (or `Remove-Item -Recurse -Force`) targets only the module's own package path inside the working folder — never the host's `src/`, `target/`, `node_modules/`, `build/`, or `dist/` at the project root. Only `prepare_environment` owns the build-output directory's lifecycle for its `.tmp/` working folder
+7. **Scope destructive operations narrowly.** Any `rm -rf` (or `Remove-Item -Recurse -Force`) targets only the module's own package path inside the working folder — never the host's `src/`, `target/`, `node_modules/`, `build/`, or `dist/` at the project root. Only `prepare_environment` owns the build-output directory's lifecycle for its system-temp working folder
 8. **Print where you are before you `cd`.** `echo "Moving to: $DIR"` saves an hour of debugging when paths are wrong. PowerShell: `Write-Host "Moving to: $DIR"`
 
 ## 1. `prepare_environment_<lang>` — copy into the host, then compile
@@ -185,7 +187,7 @@ Hard rules:
 - **Never use a "clean and test" shortcut** (`mvn clean test`, `npm run rebuild`, …) — the cleanup belongs in `prepare_environment` and would erase whatever it installed
 - **Never run the full test suite** when scoping is possible. A per-module script that runs the entire host's tests wastes minutes on every render iteration
 
-## 3. `run_conformance_tests_<lang>` — copy into `.tmp/`, then run the external suite
+## 3. `run_conformance_tests_<lang>` — copy into the system temp directory, then run the external suite
 
 Receives two positional arguments: the renderer's build output folder (`$1`) and the conformance-tests folder (`$2`).
 
@@ -193,8 +195,8 @@ Purpose: run the conformance suite in `$2` against the build that `prepare_envir
 
 Required steps:
 
-1. Validate `$1` and `$2`. **Only `$2` is actually used by the script body today** — but keep `$1` in the signature; the renderer passes both positionally
-2. Stage `$2/*` into a scratch directory under `.tmp/<lang>_conformance/`. Wipe it first if it exists. Use a deletion form that handles hidden files and odd shells safely:
+1. Validate `$1` and `$2`. **Only `$2` is actually used by the script body today** — but keep `$1` in the signature; the renderer passes both positionally. Resolve `$2` to an absolute path if it isn't one already (newer renderers pass it absolute; older ones passed it relative to the invocation directory)
+2. Stage `$2/*` into a scratch directory in the system temp directory (`/tmp/<lang>_conformance/` in Bash, the `GetTempPath()` equivalent in PowerShell). Wipe it first if it exists, and remove it again on exit (`trap 'rm -rf "$DIR"' EXIT` / `finally`). Use a deletion form that handles hidden files and odd shells safely:
    - Bash: `find "$DIR" -mindepth 1 -exec rm -rf {} +` (safer than `rm -rf "$DIR"/*`)
    - PowerShell: `Remove-Item -Recurse -Force "$DIR\*"` after confirming the path exists
 3. `cd` into the scratch directory
@@ -271,7 +273,7 @@ A working three-script set for a Java / Maven embedded integration lives under [
 
 - [`prepare_environment_java.sh`](examples/integration-embedded-testing/prepare_environment_java.sh) — copies `$1` into the host's source tree, cleans `target/`, runs `mvn clean install -DskipTests`
 - [`run_unittests_java.sh`](examples/integration-embedded-testing/run_unittests_java.sh) — re-stages into the host and runs `mvn test -Dtest='<module-pkg>.**.*Test' checkstyle:check`
-- [`run_conformance_tests_java.sh`](examples/integration-embedded-testing/run_conformance_tests_java.sh) — copies `$2` into `.tmp/java_conformance/`, builds the conformance project, runs the suite, and parses the Surefire summary line per the strict pass criteria above
+- [`run_conformance_tests_java.sh`](examples/integration-embedded-testing/run_conformance_tests_java.sh) — copies `$2` into `/tmp/java_conformance/` (system temp, removed again on exit), builds the conformance project, runs the suite, and parses the Surefire summary line per the strict pass criteria above
 
 Use them as a template when adding a new embedded integration in Java / Maven — copy, search-and-replace only the package segment and the `-Dtest` filter, then wire into `config.yaml`. For other languages (Python, Node.js, Go, Rust, …), the `implement-*-testing-script` skills generate the equivalent three-script set following the same contract.
 

package/forge/rules/integration-embedded.md

@@ -10,7 +10,7 @@ When an integration `.plain` module is **embedded** — meaning the generated co
 
 Embedded means: the host codebase already exists, has its own language / framework / dependency manager / packaging layout, and the integration must conform to all of that without negotiation.
 
-> **For test-script authoring**, also follow [`integration-embedded-testing.md`](integration-embedded-testing.md). It defines the per-script contract (`prepare_environment_<lang>`, `run_unittests_<lang>`, `run_conformance_tests_<lang>`) — staging into the host vs `.tmp/`, arg validation, exit codes, output parsing, the three `***implementation reqs***` entries the spec must declare so the scripts can be generated. This file (`integration-embedded.md`) only summarizes the test-script wiring; the testing rule is the source of truth.
+> **For test-script authoring**, also follow [`integration-embedded-testing.md`](integration-embedded-testing.md). It defines the per-script contract (`prepare_environment_<lang>`, `run_unittests_<lang>`, `run_conformance_tests_<lang>`) — staging into the host vs the system temp directory, arg validation, exit codes, output parsing, the three `***implementation reqs***` entries the spec must declare so the scripts can be generated. This file (`integration-embedded.md`) only summarizes the test-script wiring; the testing rule is the source of truth.
 
 ## The host codebase dictates the tech stack (hard rule)
 
@@ -119,7 +119,7 @@ The renderer reads the directives from the spec and the shapes from the linked s
 
 ## Test-script wiring — copy into the host, run tests there
 
-Embedded integrations are tested **inside the host codebase itself**. The prepare and unit-test scripts copy the renderer's output (`$1`, i.e. `plain_modules/<module>/`) into the host's source tree at the module's package path, then compile / test the host project in place. Only the conformance script uses a `.tmp/` scratch folder, because the conformance suite is a separate project that consumes the host build as a dependency.
+Embedded integrations are tested **inside the host codebase itself**. The prepare and unit-test scripts copy the renderer's output (`$1`, i.e. `plain_modules/<module>/`) into the host's source tree at the module's package path, then compile / test the host project in place. Only the conformance script uses a scratch folder in the system temp directory (`/tmp/<lang>_conformance/`), because the conformance suite is a separate project that consumes the host build as a dependency.
 
 This matters because the integration's generated code references host symbols by their full import path (e.g. `from host_project.integrations.base import IntegrationContract`). Those imports only resolve cleanly when the test process is rooted in the host's package layout — anything else creates path edge cases that bite later in conformance failures.
 
@@ -127,7 +127,7 @@ See [`integration-embedded-testing.md`](integration-embedded-testing.md) for the
 
 - **`prepare_environment_<lang>`** copies `$1` into the host's source tree at the module's package path, cleans the host's build-output directory, then runs the host's install / build (e.g. `mvn clean install -DskipTests`). The conformance suite later resolves the integration from the host's local dependency cache
 - **`run_unittests_<lang>`** repeats the same copy into the host (self-contained — must work without `prepare_environment` having run first), then runs the module's unit tests + lint scoped to the module's package
-- **`run_conformance_tests_<lang>`** copies `$2` (the conformance-tests folder) into `.tmp/<lang>_conformance/`, `cd`s in, builds the conformance project, and runs it against the build that `prepare_environment` already installed into the host
+- **`run_conformance_tests_<lang>`** copies `$2` (the conformance-tests folder) into `/tmp/<lang>_conformance/` (system temp), `cd`s in, builds the conformance project, and runs it against the build that `prepare_environment` already installed into the host
 
 ### Invariants the scripts must enforce
 
@@ -150,7 +150,7 @@ Before declaring an embedded integration done, in addition to the shared checkli
 - [ ] Host-package version pins are copied into `***implementation reqs***`
 - [ ] `prepare_environment` copies `$1` into the host's source tree at the module's package path, cleans the host's build-output directory, and runs the host's install / build so the conformance suite can resolve the integration from the local dependency cache
 - [ ] `run_unittests` runs the same copy-into-host sequence (self-contained — does not depend on `prepare_environment` having run) and invokes the host's test runner scoped to the module's package
-- [ ] `run_conformance_tests` copies `$2` into `.tmp/<lang>_conformance/`, `cd`s in, builds the conformance project, and runs it against the host build that `prepare_environment` already installed
+- [ ] `run_conformance_tests` copies `$2` into `/tmp/<lang>_conformance/` (system temp), `cd`s in, builds the conformance project, and runs it against the host build that `prepare_environment` already installed
 - [ ] Host codebase root is read from a named env var (default value documented in each script's usage) — never hardcoded; the env var name is captured in the integration's configuration concept
 - [ ] `***implementation reqs***` declares **everything about `:UnitTests:`** — integration source path, `:UnitTests:` source path, `:UnitTests:` package, framework + conventions, lint / static-analysis gate — per [`integration-embedded-testing.md`](integration-embedded-testing.md)
 - [ ] `***test reqs***` declares **everything about `:ConformanceTests:`** — source location, framework + execution command, package, mocking / network policy, pass criteria, build / install needs — per [`integration-embedded-testing.md`](integration-embedded-testing.md)

package/forge/skills/implement-conformance-testing-script/SKILL.md

@@ -27,8 +27,8 @@ Read both before writing anything — every script you produce must be a faithfu
 
 A conformance script is structurally very close to a unit-test script (see the sibling skill [`implement-unit-testing-script`](../implement-unit-testing-script/SKILL.md)) but with two important differences:
 
-1. **Two positional arguments instead of one.** A conformance script takes both the **build folder** (source under test) and a **separate conformance tests folder** (the tests to execute against that build).
-2. **Tests are loaded from outside the working folder.** The build is staged into `.tmp/<lang>_<arg>` and the script `cd`s into it, but the test command is pointed at the *original* `$current_dir/<conformance_tests_folder>`. Tests are never copied into the staging area.
+1. **Two positional arguments instead of one.** A conformance script takes both the **build folder** (source under test) and a **separate conformance tests folder** (the tests to execute against that build). Both arrive as **absolute paths** from current renderers (older renderer versions passed relative ones) — the script must work with both.
+2. **Tests are loaded from outside the working folder.** The build is staged into the working folder `<temp>/<lang>_<basename>` (system temp directory + language id + basename of `$1` — see [Conventions](#conventions)) and the script `cd`s into it, but the test command is pointed at the conformance tests folder, resolved to an **absolute path** up front (`$CONFORMANCE_TESTS_FOLDER` below). Tests are never copied into the staging area.
 
 Everything else — toolchain check, build staging, dependency isolation, exit codes — is the same.
 
@@ -50,9 +50,11 @@ That per-spec invocation pattern is what makes the install step expensive. A nai
 The two variants are a direct response to this:
 
 - **Install-inline** is correct only when N is small (a few specs) or dependencies are cheap. It is self-contained: stage, install, run, repeat from scratch every invocation.
-- **Activate-only** is the production answer. [`prepare_environment_<lang>`](../implement-prepare-environment-script/SKILL.md) runs **once per render** and pays the install cost a single time, populating `.tmp/<lang>_<arg>/` with the warmed environment. Each of the N conformance invocations then just **attaches** to that working folder and runs the tests — no install, no compile, just activate-and-go.
+- **Activate-only** is the production answer. [`prepare_environment_<lang>`](../implement-prepare-environment-script/SKILL.md) runs **once per render** and pays the install cost a single time, populating `<temp>/<lang>_<basename>/` with the warmed environment. Each of the N conformance invocations then just **attaches** to that working folder and runs the tests — no install, no compile, just activate-and-go.
 
-**Why picking the right variant matters:** if you emit the install-inline variant alongside an existing prepare script, prepare's work is wiped (by the script's `rm -rf .tmp/<lang>_$1`) or duplicated (by re-running install) on every run — defeating prepare's whole purpose. Conversely, emitting activate-only without a prepare script means the "verify prepared environment" check fails on every run because nothing has populated the working folder. See [Anti-Patterns](#anti-patterns).
+**The cleanup contract follows from this split:** the **install-inline** variant owns its working folder and removes it on exit (`trap 'rm -rf "$WORKING_FOLDER"' EXIT` / `Remove-Item` in `finally`). The **activate-only** variant **never deletes the working folder** — not on success, not on failure. Prepare owns that folder's lifecycle; if activate-only deleted it, the second of the N invocations would fail the "prepared environment missing" guard and the amortization would be destroyed.
+
+**Why picking the right variant matters:** if you emit the install-inline variant alongside an existing prepare script, prepare's work is wiped (by the script's staging `rm -rf` and its exit cleanup) or duplicated (by re-running install) on every run — defeating prepare's whole purpose. Conversely, emitting activate-only without a prepare script means the "verify prepared environment" check fails on every run because nothing has populated the working folder. See [Anti-Patterns](#anti-patterns).
 
 ## Pick the Shell First
 
@@ -73,36 +75,42 @@ Steps **1–3** and **step 8** are identical in both variants. Steps **4–7** d
 
 1. **Toolchain check.** Verify that the required language runtime / build tool (and the required version, if any) is installed. If not, print an error and exit with code `69`.
 2. **Argument validation.** Require **two** positional arguments: `<build_folder>` and `<conformance_tests_folder>`. If either is missing, print usage and exit with code `69`.
-3. **Capture original cwd.** Store `pwd` in a variable (`current_dir` / `$PWD`) **before** changing directories — the test command in step 8 needs it to resolve the conformance tests folder.
+3. **Resolve the conformance tests folder to an absolute path.** Current renderers pass `$2` as an absolute path; older ones passed it relative to the invocation directory. Resolve it once, up front, into a variable the test command in step 8 uses:
+   - Bash: `case "$2" in /*) CONFORMANCE_TESTS_FOLDER="$2" ;; *) CONFORMANCE_TESTS_FOLDER="$(pwd)/$2" ;; esac`
+   - PowerShell: `if (-not [System.IO.Path]::IsPathRooted($ConformanceTestsFolder)) { $ConformanceTestsFolder = Join-Path (Get-Location) $ConformanceTestsFolder }`
+
+   Never concatenate `$current_dir/$2` at the point of use — with an absolute `$2` that produces a doubled path (`/project//abs/path/...`), the classic symptom being `ImportError: Start directory is not importable` pointing at a path that contains the project root twice.
 
 ### Steps 4–7 — install-inline variant (no prepare script)
 
-4. **Working directory setup.** Define a working folder at `.tmp/<lang>_<arg1>`. Wipe it (`rm -rf` / `Remove-Item -Recurse -Force`) and recreate it. This folder — and **only** this folder — is where every subsequent write must land.
+4. **Working directory setup.** Define a working folder at `<temp>/<lang>_<basename>` (`/tmp/<lang>_$(basename "$1")` in Bash, `Join-Path ([System.IO.Path]::GetTempPath()) "<lang>_$(Split-Path $BuildFolder -Leaf)"` in PowerShell). Wipe it (`rm -rf` / `Remove-Item -Recurse -Force`) and recreate it. Register cleanup so the folder is removed when the script exits — `trap 'rm -rf "$WORKING_FOLDER"' EXIT` in Bash, `Remove-Item -Recurse -Force` in the `finally` block in PowerShell. This folder — and **only** this folder — is where every subsequent write must land.
 5. **Copy the build.** Recursively copy everything from `<build_folder>` (`$1`) into the working folder. **Do not** copy the conformance tests — they stay where they are. After this step both `$1` (build folder) and `$2` (conformance tests folder) are treated as **read-only** for the rest of the script.
-6. **Enter the working directory.** `cd` / `Set-Location` into `.tmp/<lang>_<arg1>`. If that fails, exit with code `69`. All remaining steps run from inside the working folder; they must never write back to `$1` or `$2`.
-7. **Install dependencies into an isolated environment inside `.tmp/<lang>_<arg1>`.** Set up a per-working-folder dependency location (a Python venv at `./.venv`, a local `./node_modules`, a project-scoped Maven repo at `./.m2`, etc.) and install/resolve all dependencies into it. **Never** install into the source build folder (`$1`), the conformance tests folder (`$2`), the user's global cache (`~/.m2`, system-wide `pip`, `~/.cargo`, `~/.npm`, ...), or anywhere outside `.tmp/<lang>_<arg1>`. If the install command fails, propagate its exit code immediately and **do not** proceed to step 8. See [Dependency isolation (install-inline)](#dependency-isolation-install-inline).
+6. **Enter the working directory.** `cd` / `Set-Location` into `<temp>/<lang>_<basename>`. If that fails, exit with code `69`. All remaining steps run from inside the working folder; they must never write back to `$1` or `$2`.
+7. **Install dependencies into an isolated environment inside `<temp>/<lang>_<basename>`.** Set up a per-working-folder dependency location (a Python venv at `./.venv`, a local `./node_modules`, a project-scoped Maven repo at `./.m2`, etc.) and install/resolve all dependencies into it. **Never** install into the source build folder (`$1`), the conformance tests folder (`$2`), the user's global cache (`~/.m2`, system-wide `pip`, `~/.cargo`, `~/.npm`, ...), or anywhere outside `<temp>/<lang>_<basename>`. If the install command fails, propagate its exit code immediately and **do not** proceed to step 8. See [Dependency isolation (install-inline)](#dependency-isolation-install-inline).
 
 ### Steps 4–7 — activate-only variant (prepare script exists)
 
 4. **Verify the prepared environment.** Both:
-   - Check that the working folder `.tmp/<lang>_<arg1>` exists.
+   - Check that the working folder `<temp>/<lang>_<basename>` exists — derived **exactly** the way prepare derives it (system temp dir + `<lang>_` + basename of `$1`).
    - Check that the language's isolation location inside it exists (e.g. `.venv/bin/activate` for Python, `.m2/` for Java, `node_modules/` for Node, `.gocache/` for Go, `.cargo/` for Rust).
 
    If either check fails, print a helpful error (`"Error: prepared environment missing — did you run prepare_environment_<lang>.<sh|ps1> first?"`) and exit `69`. **Do not silently fall back to creating it inline** — that would mask a real misconfiguration and turn this script into the install-inline variant in disguise. After this step both `$1` and `$2` are treated as **read-only** for the rest of the script.
-5. **Enter the working directory.** `cd` / `Set-Location` into `.tmp/<lang>_<arg1>`. If that fails, exit `69`. All remaining steps run from inside the working folder; they must never write back to `$1` or `$2`.
+
+   **No cleanup registration in this variant.** The working folder belongs to prepare; this script must never delete it — not via a `trap`, not in a `finally` block, not on failure. See [the cleanup contract](#variant-decision-install-inline-vs-activate-only).
+5. **Enter the working directory.** `cd` / `Set-Location` into `<temp>/<lang>_<basename>`. If that fails, exit `69`. All remaining steps run from inside the working folder; they must never write back to `$1` or `$2`.
 6. **Activate the prepared dependency environment.** Per-language:
    - Python: `source .venv/bin/activate` (must succeed; exit `69` on failure).
    - Java: set `MAVEN_LOCAL_REPO="$(pwd)/.m2"` so it can be passed as `-Dmaven.repo.local="$MAVEN_LOCAL_REPO"` to `mvn` in step 8.
    - Node.js / Go / Rust: nothing to activate explicitly — the test command in step 8 just needs to receive the same isolation flag/env var that prepare used (`./node_modules` is found by default; pass `GOMODCACHE` / `CARGO_HOME`).
 
-   Activation is **always relative to the working folder**, never to `$1` or `$2` — prepare populated `.tmp/<lang>_<arg1>/...`, and that is the only place to attach to.
+   Activation is **always relative to the working folder**, never to `$1` or `$2` — prepare populated `<temp>/<lang>_<basename>/...`, and that is the only place to attach to.
 7. *(There is no step 7 in this variant — install was prepare's job. Skip straight to step 8.)*
 
 ### Common step 8 (both variants)
 
-8. **Run the conformance tests.** Invoke the language's standard test command, **pointed at `$current_dir/<conformance_tests_folder>`** (the original cwd from step 3 + the second arg). The script's final exit code is whatever the test command returns — except for the "no tests discovered" case below.
+8. **Run the conformance tests.** Invoke the language's standard test command, **pointed at `$CONFORMANCE_TESTS_FOLDER`** (the absolute path resolved in step 3). The script's final exit code is whatever the test command returns — except for the "no tests discovered" case below.
 
-   The test command is **read-only** with respect to `$current_dir/$2`. It loads test files from there, but any artifacts the runner produces (caches, JUnit XML, coverage reports, compiled test classes, etc.) must land inside `.tmp/<lang>_<arg1>`, not next to the test files. If your chosen runner defaults to writing output beside the tests, pass an explicit output-directory flag pointing inside the working folder (e.g. `pytest --basetemp=./.pytest_tmp`, `jest --cacheDirectory=./.jest_cache`, Maven `target/` under `.tmp` via `mvn -f "$current_dir/$2/pom.xml" -Dproject.build.directory="$(pwd)/target" test`).
+   The test command is **read-only** with respect to `$CONFORMANCE_TESTS_FOLDER`. It loads test files from there, but any artifacts the runner produces (caches, JUnit XML, coverage reports, compiled test classes, etc.) must land inside `<temp>/<lang>_<basename>`, not next to the test files. If your chosen runner defaults to writing output beside the tests, pass an explicit output-directory flag pointing inside the working folder (e.g. `pytest --basetemp=./.pytest_tmp`, `jest --cacheDirectory=./.jest_cache`, Maven `target/` under the working folder via `mvn -f "$CONFORMANCE_TESTS_FOLDER/pom.xml" -Dproject.build.directory="$(pwd)/target" test`).
 
 ### Read-only inputs — hard rule
 
@@ -110,15 +118,15 @@ A conformance script has **two** read-only inputs: the source build folder (`$1`
 
 - install dependencies into `$1` or `$2` (no `pip install` inside `$1`/`$2`, no `npm install` inside them, no `mvn install` writing into them, no Cargo build artifacts ending up under them),
 - write a virtualenv / `node_modules` / `.m2` / `.gocache` / `.cargo` directory inside `$1` or `$2`,
-- run the test command with its `cwd` set to `$1` or `$2` (every test command runs from inside `.tmp/<lang>_<arg1>` after the `cd` in step 6 / activate-only step 5),
+- run the test command with its `cwd` set to `$1` or `$2` (every test command runs from inside `<temp>/<lang>_<basename>` after the `cd` in step 6 / activate-only step 5),
 - create logs, caches, build outputs, JUnit XML, coverage reports, compiled test classes, or temp files inside `$1` or `$2`.
 
 Why each input is read-only:
 
-- **`$1` (build folder)** is shared with the renderer (`plain_modules/...` by default) and downstream tooling. Writing into it corrupts the renderer's view of "what was generated" and breaks subsequent renders. The whole point of staging into `.tmp/` is so the source folder stays a clean, reproducible artifact of the render.
+- **`$1` (build folder)** is shared with the renderer (`plain_modules/...` by default) and downstream tooling. Writing into it corrupts the renderer's view of "what was generated" and breaks subsequent renders. The whole point of staging into the system temp directory is so the source folder stays a clean, reproducible artifact of the render.
 - **`$2` (conformance tests folder)** is the user's authored test source — typically checked into version control. Writing into it pollutes the working tree, churns git status, and (with frameworks that auto-discover) can make subsequent runs pick up generated files as if they were tests.
 
-If you find yourself about to issue any command whose `cwd` is `$1` or `$2`, or whose target path starts with `$1/` or `$2/`, **stop**. Either move the operation into `.tmp/<lang>_<arg1>`, or you're doing something the script must not do.
+If you find yourself about to issue any command whose `cwd` is `$1` or `$2`, or whose target path starts with `$1/` or `$2/`, **stop**. Either move the operation into `<temp>/<lang>_<basename>`, or you're doing something the script must not do.
 
 ### "No tests discovered" detection
 
@@ -140,19 +148,20 @@ Shared across both shell flavors **and** both variants:
   - `69` — unrecoverable invocation error: missing argument, missing toolchain, can't enter working folder, can't create venv (install-inline), or prepared environment missing/broken (activate-only). Matches the reference scripts' `UNRECOVERABLE_ERROR_EXIT_CODE`.
   - `1` — "no tests discovered" guard tripped (see above).
   - Any other non-zero code — propagated from the underlying test command.
-- **Working folder naming:** `.tmp/<lang>_<arg1>` where `<lang>` is a short identifier for the language (`java`, `python`, `node`, `go`, `rust`, ...). Use the *first* argument (the build folder) in the path, never the conformance tests folder. All dependency installs, build outputs, caches, test runner artifacts, and the test invocation itself live inside this folder. Nothing the script does should touch `$1` after step 5 (install-inline) / step 4 (activate-only), or `$2` at any point.
+- **Working folder naming:** `<temp>/<lang>_<basename>` — the system temp directory (`/tmp` in Bash, `[System.IO.Path]::GetTempPath()` in PowerShell) plus a short identifier for the language (`java`, `python`, `node`, `go`, `rust`, ...) and the **basename** of the *first* argument (the build folder — `$(basename "$1")` / `Split-Path $BuildFolder -Leaf`), never the conformance tests folder and never the raw argument (it may be absolute; concatenating it produces broken doubled paths). All dependency installs, build outputs, caches, test runner artifacts, and the test invocation itself live inside this folder. Nothing the script does should touch `$1` after step 5 (install-inline) / step 4 (activate-only), or `$2` at any point.
+- **Cleanup ownership:** the **install-inline** variant removes its working folder on exit, success and failure alike (`trap 'rm -rf "$WORKING_FOLDER"' EXIT` in Bash; `Remove-Item -Recurse -Force` in the PowerShell `finally` block, after `Pop-Location`). The **activate-only** variant **never deletes the working folder** — prepare owns its lifecycle and re-creates it once per render; deleting it here would break every subsequent per-spec invocation.
 - **Logging — be as verbose as possible.** The script is the only thing the operator sees between a `codeplain` render and a green/red conformance result; when it fails, the only forensic evidence anyone has is its stdout/stderr. Treat the script like a production runbook, not a quiet helper. Concretely:
-  - **Announce every step before doing it**, with the exact value of every variable involved — the resolved build folder, the resolved conformance-tests folder, the captured `current_dir`, the working folder, the language, the toolchain version detected, the isolation root being activated (install-inline) or verified (activate-only), the variant in use, and the test command about to run. "Running tests" alone is useless; "Running `python -m unittest discover -b -s /abs/path/to/tests` from working folder `.tmp/python_build/` with venv activated at `./.venv` (Python 3.11.6)" is triage-ready.
+  - **Announce every step before doing it**, with the exact value of every variable involved — the resolved build folder, the resolved conformance-tests folder, the working folder, the language, the toolchain version detected, the isolation root being activated (install-inline) or verified (activate-only), the variant in use, and the test command about to run. "Running tests" alone is useless; "Running `python -m unittest discover -b -s /abs/path/to/tests` from working folder `/tmp/python_build/` with venv activated at `./.venv` (Python 3.11.6)" is triage-ready.
   - **Echo every non-trivial command before executing it.** In Bash, use `set -x` for the body of the script (or `echo "+ <cmd>"` immediately before each call); in PowerShell, set `$VerbosePreference = 'Continue'` and use `Write-Verbose` / `Write-Host` to print each command line with its arguments.
   - **Print clear section banners.** Each pattern step gets a banner like `===== [6/8] Activate prepared environment =====` (activate-only) or `===== [7/8] Install dependencies =====` (install-inline) so a long log can be navigated by eye, **and** the banner names the variant so the operator immediately knows which branch is executing.
-  - **Print resolved absolute paths**, not just the relative names. Operators reading the log on a different machine need to know where the venv / `.m2` / `node_modules` / `.gocache` / `.cargo` actually landed, and where `$2` resolved to relative to `current_dir`.
+  - **Print resolved absolute paths**, not just the relative names. Operators reading the log on a different machine need to know where the venv / `.m2` / `node_modules` / `.gocache` / `.cargo` actually landed, and what `$2` resolved to.
   - **Print the toolchain's own `--version` output** (and the path it resolved from) during step 1, even on success. This is the single most useful piece of forensic data when conformance passes locally but fails in CI.
   - **Surface the install / activation output verbatim** — do not pipe it to `/dev/null`, do not redirect to a log file inside the working folder. The operator must see every dependency-resolver line (install-inline) or every "venv missing" / "\.m2 missing" guard message (activate-only) in the script's own output stream. Drop the `VERBOSE`-gated wrapping from the Python reference — conformance output is where forensic data is most valuable.
-  - **Print which variant the script is** (install-inline vs. activate-only) and **why** — e.g. "activate-only variant: detected `.tmp/python_build/.venv/bin/activate` from a prior `prepare_environment_python.sh` run". When the verify step trips, the operator must see both the path that was checked and the expected source ("did you run `prepare_environment_python.sh <build>` first?").
-  - **On failure, print what was about to run before exiting** — the exact test command, the working directory, the captured `current_dir`, the resolved `$2`, and the relevant environment variables (`PYTHONPATH`, `NODE_PATH`, `GOMODCACHE`, `CARGO_HOME`, `MAVEN_OPTS`, `PATH`, plus `VIRTUAL_ENV` for Python).
-  - **Print a final summary line** that names the variant, the test command, the exit code, the captured `current_dir`, and the working folder so the operator knows exactly what to re-run by hand if needed.
+  - **Print which variant the script is** (install-inline vs. activate-only) and **why** — e.g. "activate-only variant: detected `/tmp/python_build/.venv/bin/activate` from a prior `prepare_environment_python.sh` run". When the verify step trips, the operator must see both the path that was checked and the expected source ("did you run `prepare_environment_python.sh <build>` first?").
+  - **On failure, print what was about to run before exiting** — the exact test command, the working directory, the resolved `$CONFORMANCE_TESTS_FOLDER`, and the relevant environment variables (`PYTHONPATH`, `NODE_PATH`, `GOMODCACHE`, `CARGO_HOME`, `MAVEN_OPTS`, `PATH`, plus `VIRTUAL_ENV` for Python).
+  - **Print a final summary line** that names the variant, the test command, the exit code, the resolved conformance-tests folder, and the working folder so the operator knows exactly what to re-run by hand if needed.
   Verbosity is not noise here — a chatty script that documents itself in its own output is the difference between a 30-second triage and a 30-minute one. Never trade verbosity for terseness.
-- **Capture `current_dir` before `cd`.** This is the single most common bug in hand-written conformance scripts: forgetting that the conformance tests folder argument is relative to the *invocation* directory, not the working folder. The captured value must also appear in the verbose log (per the logging rule above) so misresolutions of `$2` are visible at a glance.
+- **Resolve `$2` before any `cd`.** This is the single most common bug in hand-written conformance scripts: using the conformance-tests argument after changing directory. Resolve it to an absolute path in step 3 (absolute → use as-is; relative → prefix the invocation directory) and use only the resolved variable afterwards. The resolved value must also appear in the verbose log (per the logging rule above) so misresolutions of `$2` are visible at a glance.
 
 ### Dependency isolation (install-inline)
 
@@ -160,17 +169,17 @@ This section applies to **install-inline scripts only.** For activate-only scrip
 
 The dependency environment must live **inside** `$WORKING_FOLDER` so the test run can't be polluted by — or pollute — the user's global caches. Pick the most idiomatic isolation mechanism for the language:
 
-| Language | Isolation mechanism | Install command (run inside `$WORKING_FOLDER`) | Test command (point at `$current_dir/$2`) |
+| Language | Isolation mechanism | Install command (run inside `$WORKING_FOLDER`) | Test command (point at `$CONFORMANCE_TESTS_FOLDER`) |
 |---|---|---|---|
-| Python | `venv` at `./.venv` | `python3 -m venv .venv && source .venv/bin/activate && pip install -r requirements.txt` | `python -m unittest discover -b -s "$current_dir/$2"` (or `pytest "$current_dir/$2"`) |
-| Node.js | local `./node_modules` (default) | `npm ci` (preferred) or `npm install` | `npx jest --rootDir "$current_dir/$2"` |
-| Java | project-scoped Maven repo at `./.m2` | `mvn -Dmaven.repo.local=./.m2 install -DskipTests` (build + install artifact so the test pom can resolve it) | `mvn -f "$current_dir/$2/pom.xml" -Dmaven.repo.local="$(pwd)/.m2" test` |
-| Go | module cache at `./.gocache` | `GOMODCACHE="$PWD/.gocache" go mod download` (optional pre-warm) | `GOMODCACHE="$PWD/.gocache" go test "$current_dir/$2/..."` |
-| Rust | cargo home at `./.cargo` | `CARGO_HOME="$PWD/.cargo" cargo fetch` (optional pre-warm) | `CARGO_HOME="$PWD/.cargo" cargo test --manifest-path "$current_dir/$2/Cargo.toml"` |
+| Python | `venv` at `./.venv` | `python3 -m venv .venv && source .venv/bin/activate && pip install -r requirements.txt` | `python -m unittest discover -b -s "$CONFORMANCE_TESTS_FOLDER"` (or `pytest "$CONFORMANCE_TESTS_FOLDER"`) |
+| Node.js | local `./node_modules` (default) | `npm ci` (preferred) or `npm install` | `npx jest --rootDir "$CONFORMANCE_TESTS_FOLDER"` |
+| Java | project-scoped Maven repo at `./.m2` | `mvn -Dmaven.repo.local=./.m2 install -DskipTests` (build + install artifact so the test pom can resolve it) | `mvn -f "$CONFORMANCE_TESTS_FOLDER/pom.xml" -Dmaven.repo.local="$(pwd)/.m2" test` |
+| Go | module cache at `./.gocache` | `GOMODCACHE="$PWD/.gocache" go mod download` (optional pre-warm) | `GOMODCACHE="$PWD/.gocache" go test "$CONFORMANCE_TESTS_FOLDER/..."` |
+| Rust | cargo home at `./.cargo` | `CARGO_HOME="$PWD/.cargo" cargo fetch` (optional pre-warm) | `CARGO_HOME="$PWD/.cargo" cargo test --manifest-path "$CONFORMANCE_TESTS_FOLDER/Cargo.toml"` |
 
 Notes:
 
-- **Every path in the install command and test command is relative to `.tmp/<lang>_<arg1>`.** That's why the script `cd`s into the working folder in step 6 — from that point on, `./.venv`, `./node_modules`, `./.m2`, etc. all resolve under `.tmp/<lang>_<arg1>`, never under `$1` or `$2`.
+- **Every path in the install command and test command is relative to `<temp>/<lang>_<basename>`.** That's why the script `cd`s into the working folder in step 6 — from that point on, `./.venv`, `./node_modules`, `./.m2`, etc. all resolve under `<temp>/<lang>_<basename>`, never under `$1` or `$2`.
 - **Always pass the isolation flag/env var to both the install command and the test command.** They must agree on where deps live, otherwise the test command will silently fall back to the global cache **or** (worse) write into `$1` / `$2`.
 - **Python is the only ecosystem where the venv is mandatory** to satisfy "into a virtual environment" literally. The others use language-native equivalents that achieve the same isolation.
 - **Propagate the install exit code immediately.** In Bash: `<install cmd> || exit $?`. In PowerShell: check `$LASTEXITCODE` and `exit $LASTEXITCODE` if non-zero.
@@ -180,13 +189,15 @@ Notes:
 
 This section applies to **activate-only scripts only.** The isolation location was created by prepare; conformance just needs to attach to it and pass the right flags to the test command.
 
-| Language | Verify exists in step 4 | Activate in step 6 | Test command in step 8 (point at `$current_dir/$2`) |
+In the table below, `$WORKING_FOLDER` is `<temp>/<lang>_<basename>` — e.g. `/tmp/python_$(basename "$1")` in Bash.
+
+| Language | Verify exists in step 4 | Activate in step 6 | Test command in step 8 (point at `$CONFORMANCE_TESTS_FOLDER`) |
 |---|---|---|---|
-| Python | `.tmp/<lang>_$1/.venv/bin/activate` | `source .venv/bin/activate` (after `cd`-ing into the working folder) | `python -m unittest discover -b -s "$current_dir/$2"` |
-| Node.js | `.tmp/<lang>_$1/node_modules/` | (nothing) | `npx jest --rootDir "$current_dir/$2"` |
-| Java | `.tmp/<lang>_$1/.m2/` | `MAVEN_LOCAL_REPO="$(pwd)/.m2"` | `mvn -f "$current_dir/$2/pom.xml" -Dmaven.repo.local="$MAVEN_LOCAL_REPO" test` |
-| Go | `.tmp/<lang>_$1/.gocache/` | `export GOMODCACHE="$(pwd)/.gocache"` | `go test "$current_dir/$2/..."` |
-| Rust | `.tmp/<lang>_$1/.cargo/` | `export CARGO_HOME="$(pwd)/.cargo"` | `cargo test --manifest-path "$current_dir/$2/Cargo.toml"` |
+| Python | `$WORKING_FOLDER/.venv/bin/activate` | `source .venv/bin/activate` (after `cd`-ing into the working folder) | `python -m unittest discover -b -s "$CONFORMANCE_TESTS_FOLDER"` |
+| Node.js | `$WORKING_FOLDER/node_modules/` | (nothing) | `npx jest --rootDir "$CONFORMANCE_TESTS_FOLDER"` |
+| Java | `$WORKING_FOLDER/.m2/` | `MAVEN_LOCAL_REPO="$(pwd)/.m2"` | `mvn -f "$CONFORMANCE_TESTS_FOLDER/pom.xml" -Dmaven.repo.local="$MAVEN_LOCAL_REPO" test` |
+| Go | `$WORKING_FOLDER/.gocache/` | `export GOMODCACHE="$(pwd)/.gocache"` | `go test "$CONFORMANCE_TESTS_FOLDER/..."` |
+| Rust | `$WORKING_FOLDER/.cargo/` | `export CARGO_HOME="$(pwd)/.cargo"` | `cargo test --manifest-path "$CONFORMANCE_TESTS_FOLDER/Cargo.toml"` |
 
 Notes:
 
@@ -209,7 +220,7 @@ Notes:
 - **Exit codes:** use `exit 69` etc. (PowerShell honors them just like Bash).
 - **Toolchain check:** prefer `Get-Command <tool> -ErrorAction SilentlyContinue` and, where a specific version is needed, parse the tool's `--version` output.
 - **Filesystem:** use `Test-Path`, `Remove-Item -Recurse -Force`, `New-Item -ItemType Directory`, `Copy-Item -Recurse`, `Set-Location`. Quote paths to handle spaces.
-- **Capture original cwd:** `$currentDir = (Get-Location).Path` **before** any `Set-Location` call.
+- **Resolve `$ConformanceTestsFolder` to absolute** with `[System.IO.Path]::IsPathRooted` + `Join-Path (Get-Location) ...` **before** any `Set-Location` call.
 - **No `chmod` step needed.** If execution policy is likely to block the script, mention `Set-ExecutionPolicy -Scope CurrentUser RemoteSigned` to the user — don't bake it into the script.
 
 ## Workflow
@@ -229,19 +240,21 @@ Notes:
 
 ## Anti-Patterns
 
-- **(Hard mistake) Don't install into, build into, or otherwise write to the source build folder (`$1`) or the conformance tests folder (`$2`).** Both arguments are read-only input. Every install, cache, build artifact, log, JUnit XML, coverage report, compiled test class, and temp file must land in `.tmp/<lang>_<arg1>`. This includes never running `pip install`, `npm install`, `mvn install`, or `cargo build` with `$1` or `$2` as their `cwd` or target; never letting a venv / `node_modules` / `.m2` / `.gocache` / `.cargo` directory appear inside `$1` or `$2`; and never running the test command from inside either folder. The whole point of staging into `.tmp/` is so the build folder remains a clean artifact of the render and the conformance tests folder remains a clean tree under the user's version control — writing to either one corrupts those guarantees.
-- **Don't emit the install-inline variant when a `prepare_environment_<lang>` script already exists.** The conformance script's `rm -rf .tmp/<lang>_$1` will wipe everything prepare did, and the inline install will redo it from scratch on every run. Always run the [Variant decision](#variant-decision-install-inline-vs-activate-only) check first.
+- **(Hard mistake) Don't install into, build into, or otherwise write to the source build folder (`$1`) or the conformance tests folder (`$2`).** Both arguments are read-only input. Every install, cache, build artifact, log, JUnit XML, coverage report, compiled test class, and temp file must land in `<temp>/<lang>_<basename>`. This includes never running `pip install`, `npm install`, `mvn install`, or `cargo build` with `$1` or `$2` as their `cwd` or target; never letting a venv / `node_modules` / `.m2` / `.gocache` / `.cargo` directory appear inside `$1` or `$2`; and never running the test command from inside either folder. The whole point of staging into the system temp directory is so the build folder remains a clean artifact of the render and the conformance tests folder remains a clean tree under the user's version control — writing to either one corrupts those guarantees.
+- **(Hard mistake) Don't build any path by concatenating a raw argument.** `$1` and `$2` arrive as absolute paths from current renderers. Patterns like `WORKING_FOLDER=.tmp/$1`, `WORKING_FOLDER=<lang>_$1`, or `"$current_dir/$2"` silently produce doubled paths (`/project//abs/path/...`) — the classic symptom is `ImportError: Start directory is not importable` or "folder does not exist" pointing at a path that contains the project root twice. Always derive the working folder from `$(basename "$1")` / `Split-Path -Leaf`, and use the `$CONFORMANCE_TESTS_FOLDER` resolved in step 3.
+- **(Hard mistake — activate-only) Don't delete the working folder.** No `trap 'rm -rf …' EXIT`, no `Remove-Item` in `finally`, no cleanup on failure. The working folder is prepare's deliverable and is attached to N times per render; deleting it after the first invocation makes every later invocation fail the "prepared environment missing" guard. Cleanup belongs to the install-inline variant only.
+- **Don't emit the install-inline variant when a `prepare_environment_<lang>` script already exists.** The conformance script's staging `rm -rf` and exit cleanup will wipe everything prepare did, and the inline install will redo it from scratch on every run. Always run the [Variant decision](#variant-decision-install-inline-vs-activate-only) check first.
 - **Don't emit the activate-only variant when no prepare script exists.** The "verify prepared environment" check will fail on every run because nothing has populated the working folder.
 - **Don't silently fall back from activate-only to install-inline** when the prepared environment is missing. Exit `69` with a clear error so the misconfiguration is visible. Silent fallback hides the real bug and produces inconsistent behavior between runs.
-- **Don't copy the conformance tests folder into `.tmp/`.** Only the build folder is staged (and only in install-inline). The test folder is read in place from `$current_dir/$2`.
-- **Don't compute the test path after `cd`.** Capture `current_dir` first; otherwise `$2` will be resolved relative to the working folder and silently miss the tests.
+- **Don't copy the conformance tests folder into the working folder.** Only the build folder is staged (and only in install-inline). The test folder is read in place from `$CONFORMANCE_TESTS_FOLDER`.
+- **Don't compute the test path after `cd`.** Resolve `$2` to an absolute path in step 3; otherwise a relative `$2` will be resolved against the working folder and silently miss the tests.
 - **Don't skip the "no tests discovered" check.** A conformance suite that finds zero tests and exits `0` is the worst possible failure mode — it looks like success in CI.
 - **Don't skip the toolchain check**, even when "everyone has it installed" — exit code `69` is what the calling system relies on to detect a missing runtime.
-- **Don't reuse the source folder in place** (install-inline). Always copy into `.tmp/<lang>_<arg1>` first; the renderer relies on this isolation.
+- **Don't reuse the source folder in place** (install-inline). Always copy into `<temp>/<lang>_<basename>` first; the renderer relies on this isolation.
 - **Don't change the exit-code contract.** Other parts of the system branch on `69` and `1` specifically — and these codes must be identical between the Bash and PowerShell variants.
 - **Don't write a cross-shell hybrid** (e.g. a `.sh` that detects PowerShell, or vice versa). Ship one script per shell, named with the appropriate extension.
 - **Don't install dependencies into the user's global location** (`~/.m2`, system-wide `pip`, `~/.cargo`, etc.) in the install-inline variant. Always isolate inside `$WORKING_FOLDER` so concurrent runs and other projects can't interfere.
 - **Don't run the test command without first verifying the install / activation succeeded.** A failed install (or missing prepared env) followed by a "test" run produces misleading errors that look like test failures.
 - **Don't forget to update `config.yaml`.** After creating the conformance test script, always add or update the `conformance-tests-script:` key in `config.yaml` to reference the new script. Without this entry, the `codeplain` renderer won't know where to find the conformance test script.
 - **Don't forget to link the script as a resource in the base `.plain` files.** A script that is referenced only from `config.yaml` is invisible to the spec contract — the renderer treats `conformance-tests-script:` as a build-system pointer, not as part of the test-req contract the model reads. Use the `add-resource` skill to add a markdown link to the script from the `.plain` module that owns the conformance test reqs (see Workflow step 9). Omitting the linked resource means the model authors and reviews conformance-test code without ever seeing the actual runner it has to satisfy — and for activate-only scripts, the model also loses sight of which isolation paths must already be in place by the time tests run.
-- **Don't write a terse script.** Silent steps, `>/dev/null` redirects on the install / activation output, missing `--version` prints, absent variant banners, an un-logged `current_dir`, and a missing final summary all make the script harder to debug than the tests it is running. Follow the *Logging — be as verbose as possible* rule under [Conventions](#conventions) literally: every step announces itself, every command is echoed, the variant is named in its banner, `current_dir` and the resolved `$2` are printed, every failure prints what was about to run and where, and the final summary names the variant, test command, exit code, and working folder.
+- **Don't write a terse script.** Silent steps, `>/dev/null` redirects on the install / activation output, missing `--version` prints, absent variant banners, an un-logged resolved `$2`, and a missing final summary all make the script harder to debug than the tests it is running. Follow the *Logging — be as verbose as possible* rule under [Conventions](#conventions) literally: every step announces itself, every command is echoed, the variant is named in its banner, the resolved `$CONFORMANCE_TESTS_FOLDER` is printed, every failure prints what was about to run and where, and the final summary names the variant, test command, exit code, and working folder.

package/forge/skills/implement-conformance-testing-script/assets/run_conformance_tests_cypress.ps1

@@ -69,6 +69,17 @@ function Cleanup {
     if ($script:build_output -and (Test-Path $script:build_output)) {
         Remove-Item $script:build_output -Force -ErrorAction SilentlyContinue
     }
+
+    # Step out of any temp subfolder we may still be in before deleting it
+    Set-Location -Path ([System.IO.Path]::GetTempPath()) -ErrorAction SilentlyContinue
+
+    # Remove temporary build subfolders if they exist
+    if ($script:NODE_SUBFOLDER -and (Test-Path $script:NODE_SUBFOLDER)) {
+        Remove-Item -Path $script:NODE_SUBFOLDER -Recurse -Force -ErrorAction SilentlyContinue
+    }
+    if ($script:NODE_CONFORMANCE_TESTS_SUBFOLDER -and (Test-Path $script:NODE_CONFORMANCE_TESTS_SUBFOLDER)) {
+        Remove-Item -Path $script:NODE_CONFORMANCE_TESTS_SUBFOLDER -Recurse -Force -ErrorAction SilentlyContinue
+    }
 }
 
 # Check for and kill any existing Node server from previous runs
@@ -98,20 +109,21 @@ if ($args[2] -eq "-v" -or $args[2] -eq "--verbose") {
 $current_dir = Get-Location
 
 try {
-    # Define the path to the subfolder
-    $NODE_SUBFOLDER = "node_$BuildFolder"
+    # Define the path to the subfolder. It lives in the system temp directory;
+    # $BuildFolder may be an absolute path, so only its leaf name is used.
+    $script:NODE_SUBFOLDER = Join-Path ([System.IO.Path]::GetTempPath()) "node_$(Split-Path $BuildFolder -Leaf)"
 
     # Running React application
-    Write-Host "### Step 1: Starting the React application in folder $NODE_SUBFOLDER..."
+    Write-Host "### Step 1: Starting the React application in folder $script:NODE_SUBFOLDER..."
 
     if ($env:VERBOSE -eq "1") {
-        Write-Host "Preparing Node subfolder: $NODE_SUBFOLDER"
+        Write-Host "Preparing Node subfolder: $script:NODE_SUBFOLDER"
     }
 
     # Check if the node subfolder exists
-    if (Test-Path $NODE_SUBFOLDER) {
+    if (Test-Path $script:NODE_SUBFOLDER) {
         # Delete all files and folders except "node_modules", "plain_modules", and "package-lock.json"
-        Get-ChildItem -Path $NODE_SUBFOLDER -Force |
+        Get-ChildItem -Path $script:NODE_SUBFOLDER -Force |
             Where-Object {
                 $_.Name -ne "node_modules" -and
                 $_.Name -ne "plain_modules" -and
@@ -126,18 +138,18 @@ try {
             Write-Host "Subfolder does not exist. Creating it..."
         }
 
-        New-Item -ItemType Directory -Path $NODE_SUBFOLDER -Force | Out-Null
+        New-Item -ItemType Directory -Path $script:NODE_SUBFOLDER -Force | Out-Null
     }
 
-    Copy-Item -Path "$BuildFolder/*" -Destination $NODE_SUBFOLDER -Recurse -Force
+    Copy-Item -Path "$BuildFolder/*" -Destination $script:NODE_SUBFOLDER -Recurse -Force
 
     # Move to the subfolder
-    if (-not (Test-Path $NODE_SUBFOLDER)) {
-        Write-Host "Error: Node build folder '$NODE_SUBFOLDER' does not exist."
+    if (-not (Test-Path $script:NODE_SUBFOLDER)) {
+        Write-Host "Error: Node build folder '$script:NODE_SUBFOLDER' does not exist."
         exit $UNRECOVERABLE_ERROR_EXIT_CODE
     }
 
-    Push-Location $NODE_SUBFOLDER
+    Push-Location $script:NODE_SUBFOLDER
 
     # Temporarily allow stderr output without throwing (npm may write warnings to stderr)
     # ForEach-Object converts ErrorRecord objects (from stderr) to plain strings to avoid verbose error formatting
@@ -245,17 +257,18 @@ try {
     # Move back to the original directory
     Set-Location $current_dir
 
-    # Define the path to the conformance tests subfolder
-    $NODE_CONFORMANCE_TESTS_SUBFOLDER = "node_$ConformanceTestsFolder"
+    # Define the path to the conformance tests subfolder. It lives in the system
+    # temp directory; $ConformanceTestsFolder may be absolute, so only its leaf name is used.
+    $script:NODE_CONFORMANCE_TESTS_SUBFOLDER = Join-Path ([System.IO.Path]::GetTempPath()) "node_$(Split-Path $ConformanceTestsFolder -Leaf)"
 
     if ($env:VERBOSE -eq "1") {
-        Write-Host "Preparing conformance tests Node subfolder: $NODE_CONFORMANCE_TESTS_SUBFOLDER"
+        Write-Host "Preparing conformance tests Node subfolder: $script:NODE_CONFORMANCE_TESTS_SUBFOLDER"
     }
 
     # Check if the conformance tests node subfolder exists
-    if (Test-Path $NODE_CONFORMANCE_TESTS_SUBFOLDER) {
+    if (Test-Path $script:NODE_CONFORMANCE_TESTS_SUBFOLDER) {
         # Delete all files and folders except "node_modules", "plain_modules", and "package-lock.json"
-        Get-ChildItem -Path $NODE_CONFORMANCE_TESTS_SUBFOLDER -Force |
+        Get-ChildItem -Path $script:NODE_CONFORMANCE_TESTS_SUBFOLDER -Force |
             Where-Object {
                 $_.Name -ne "node_modules" -and
                 $_.Name -ne "plain_modules" -and
@@ -270,18 +283,18 @@ try {
             Write-Host "Subfolder does not exist. Creating it..."
         }
 
-        New-Item -ItemType Directory -Path $NODE_CONFORMANCE_TESTS_SUBFOLDER -Force | Out-Null
+        New-Item -ItemType Directory -Path $script:NODE_CONFORMANCE_TESTS_SUBFOLDER -Force | Out-Null
     }
 
-    Copy-Item -Path "$ConformanceTestsFolder/*" -Destination $NODE_CONFORMANCE_TESTS_SUBFOLDER -Recurse -Force
+    Copy-Item -Path "$ConformanceTestsFolder/*" -Destination $script:NODE_CONFORMANCE_TESTS_SUBFOLDER -Recurse -Force
 
     # Move to the subfolder with Cypress tests
-    if (-not (Test-Path $NODE_CONFORMANCE_TESTS_SUBFOLDER)) {
-        Write-Host "Error: conformance tests Node folder '$NODE_CONFORMANCE_TESTS_SUBFOLDER' does not exist."
+    if (-not (Test-Path $script:NODE_CONFORMANCE_TESTS_SUBFOLDER)) {
+        Write-Host "Error: conformance tests Node folder '$script:NODE_CONFORMANCE_TESTS_SUBFOLDER' does not exist."
         exit $UNRECOVERABLE_ERROR_EXIT_CODE
     }
 
-    Push-Location $NODE_CONFORMANCE_TESTS_SUBFOLDER
+    Push-Location $script:NODE_CONFORMANCE_TESTS_SUBFOLDER
 
     # Temporarily allow stderr output without throwing (npm may write warnings to stderr)
     # ForEach-Object converts ErrorRecord objects (from stderr) to plain strings to avoid verbose error formatting

package/forge/skills/implement-conformance-testing-script/assets/run_conformance_tests_golang.ps1

@@ -23,7 +23,15 @@ $ConformanceTestsFolder = $args[1]
 
 $current_dir = Get-Location
 
-$GO_BUILD_SUBFOLDER = "go_$BuildFolder"
+# Resolve conformance tests folder to an absolute path so it can be used
+# from the build subfolder (where we'll Push-Location to next).
+if (-not [System.IO.Path]::IsPathRooted($ConformanceTestsFolder)) {
+    $ConformanceTestsFolder = Join-Path $current_dir $ConformanceTestsFolder
+}
+
+# Working folder lives in the system temp directory. $BuildFolder may be an
+# absolute path, so only its leaf name is used to build the working folder name.
+$GO_BUILD_SUBFOLDER = Join-Path ([System.IO.Path]::GetTempPath()) "go_$(Split-Path $BuildFolder -Leaf)"
 
 if ($env:VERBOSE -eq "1") {
     Write-Host "Preparing Go build subfolder: $GO_BUILD_SUBFOLDER"
@@ -60,9 +68,9 @@ try {
     go get
 
     # Move to conformance tests folder
-    Set-Location "$current_dir/$ConformanceTestsFolder"
+    Set-Location $ConformanceTestsFolder
     if ($LASTEXITCODE -and $LASTEXITCODE -ne 0) {
-        Write-Host "Error: Conformance tests folder '$current_dir/$ConformanceTestsFolder' does not exist."
+        Write-Host "Error: Conformance tests folder '$ConformanceTestsFolder' does not exist."
         exit $UNRECOVERABLE_ERROR_EXIT_CODE
     }
 
@@ -75,7 +83,7 @@ try {
     }
 
     # Move back to build directory
-    Set-Location "$current_dir/$GO_BUILD_SUBFOLDER"
+    Set-Location $GO_BUILD_SUBFOLDER
 
     # Execute Go lang conformance tests
     Write-Host "Running Golang conformance tests...`n"
@@ -83,7 +91,7 @@ try {
     # Temporarily allow stderr output without throwing (Go may write to stderr)
     # ForEach-Object converts ErrorRecord objects (from stderr) to plain strings to avoid verbose error formatting
     $ErrorActionPreference = 'Continue'
-    $output = go run "$current_dir/$ConformanceTestsFolder/conformance_tests.go" 2>&1 | ForEach-Object { if ($_ -is [System.Management.Automation.ErrorRecord]) { $_.Exception.Message } else { $_ } } | Out-String
+    $output = go run (Join-Path $ConformanceTestsFolder "conformance_tests.go") 2>&1 | ForEach-Object { if ($_ -is [System.Management.Automation.ErrorRecord]) { $_.Exception.Message } else { $_ } } | Out-String
     $exit_code = $LASTEXITCODE
     $ErrorActionPreference = 'Stop'
 
@@ -97,4 +105,7 @@ try {
     exit $exit_code
 } finally {
     Pop-Location
+    if (Test-Path $GO_BUILD_SUBFOLDER) {
+        Remove-Item -Path $GO_BUILD_SUBFOLDER -Recurse -Force -ErrorAction SilentlyContinue
+    }
 }