feat(clients): models: Gemini on Vertex, Model Garden, Claude, endpoint models, media, VertexModel
One call table verified against Discovery (29 calls), live-verified read-only with an express key and a service account token, /agent/models example routes. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
This commit is contained in:
@@ -0,0 +1,29 @@
|
||||
// Route: POST /agent/models/generate (hard caps: output tokens, prompt size, project requirement)
|
||||
#include <cstdlib>
|
||||
|
||||
#include "support/GoogleRig.hpp"
|
||||
|
||||
using namespace molo::example::test;
|
||||
|
||||
int main() {
|
||||
unsetenv("GOOGLE_CLOUD_PROJECT");
|
||||
GoogleRig rig;
|
||||
Fixture f(rig.client);
|
||||
auto over = f.client->post("/agent/models/generate",
|
||||
"{\"prompt\":\"hi\",\"max_tokens\":\"70000\"}", "application/json");
|
||||
check(over.status == 422, "max_tokens over the 65536 cap is 422");
|
||||
auto words = f.client->post("/agent/models/generate",
|
||||
"{\"prompt\":\"hi\",\"max_tokens\":\"many\"}", "application/json");
|
||||
check(words.status == 422, "non numeric max_tokens is 422");
|
||||
std::string huge(1000001, 'x');
|
||||
auto big = f.client->post("/agent/models/count", "{\"prompt\":\"" + huge + "\"}", "application/json");
|
||||
check(big.status == 422, "prompt over the 1,000,000 char cap is 422");
|
||||
auto id = f.client->post("/agent/models/generate",
|
||||
"{\"prompt\":\"hi\",\"model\":\"a/b\"}", "application/json");
|
||||
check(id.status == 422, "model id with a slash is 422");
|
||||
auto claude = f.client->post("/agent/models/claude", "{\"prompt\":\"hi\"}", "application/json");
|
||||
check(claude.status == 422 && contains(claude.body, "project"),
|
||||
"Claude without GOOGLE_CLOUD_PROJECT is 422 naming the project");
|
||||
check(rig.mock->send_count() == 0, "side effect: nothing reached Google");
|
||||
return finish("agent/models/limits");
|
||||
}
|
||||
Reference in New Issue
Block a user