test(capabilities): workspace/tooling selftests + live demo fixture (#21)

- mock plumbing cases: workspace path + tools delivered (asserted from
  run-record stderr), host workspace created, absent fields = empty vars
- validation negatives: unknown tool, workspace traversal
- tasks/workspace-demo.json: pi uses bash inside the persistent demo
  workspace; host-visible proof.txt verified live

Closes #21
This commit is contained in:
2026-09-02 22:06:43 -05:00
parent 172368612c
commit b017e66e17
2 changed files with 41 additions and 0 deletions
+9
View File
@@ -0,0 +1,9 @@
{
"taskVersion": 1,
"id": "t-workspace-demo",
"prompt": "Use the bash tool to create a file named proof.txt in the current directory containing exactly the text: workspace works. Then reply with exactly: WORKSPACE_OK",
"workspace": "demo",
"capabilities": { "tools": ["bash", "read", "write"] },
"expectExact": "WORKSPACE_OK",
"timeoutSeconds": 180
}