make test-file-runner run tests in parallel

agents/test-file-runner.md

@@ -42,11 +42,12 @@ iterations requested.
 Second, call tps-report_confirmCaseCount to make sure you have counted the cases
 correctly. Describe your reasoning step by step.
 
-Lastly, for each test case, call the test-case-runner function with the path to
-the test file, and the test case index. The test case runner will run the test
-case and will generate as many iterations as were set when you upserted the tps
+Lastly, call the test-case-runner function with the path to
+the test file, and all the test case indexes that you want to run, which will
+most likely be all of them unless specifically asked to narrow the scope. The test case runner will run the test
+cases and will generate as many iterations as were set when you upserted the tps
 report. The test case runner will update the tps report automatically. ONLY CALL
-THE TEST CASE RUNNER ONCE FOR EACH TEST CASE INDEX.
+THE TEST CASE RUNNER ONCE.
 
 Once you have finished, call the resolve function with empty parameters. The
 resolve function must be called alone, never in parallel with other function

info/agents-list.md

@@ -78,4 +78,8 @@
 - `agents/test-results-summary.md` Produces a concise analysis on a TPS test
   result file.
 - `agents/imogen.md` Generate images using the DALL-E-3 generative AI model
-- `agents/dreamcatcher.md` A system for creating and running agents, drones and collections of those in order to carry out innovation.
+- `agents/dreamcatcher.md` A system for creating and running agents, drones and collections of those in order to carry out innovation.
+- `agents/reason.md` This agent wraps the wise and deep (but slow)
+  o1-preview reasoning model with a quick and responsive gpt-4o-mini model to do all its
+  admin work. This is the best choice for reasoning tasks that also need some
+  tools, and should be the default for general purpose reasoning.