Q83: a model passes an agent check before it does agent work
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
1 parent
4e232b241f
commit
665bf3cae1
35 files changed
+350
-47
No files matched your search
@@ -14,6 +14,7 @@ import { buildAgentInstructions, type TaskExecutionAgentRow } from "../../src/ma
|
||||
import { AGENT_TOOLS } from "../../src/main/services/agent-tools";
|
||||
import { goalKickoffDescription } from "../../src/main/services/goal-kickoff";
|
||||
import { CEO_FIRST_HIRES } from "../../src/main/services/org-building";
|
||||
import { answeringModelCheck } from "../helpers/model-check";
|
||||
|
||||
function agentRow(overrides: Partial<TaskExecutionAgentRow>): TaskExecutionAgentRow {
|
||||
return {
|
||||
@@ -134,7 +135,7 @@ describe("agent instruction wiring", () => {
|
||||
capturedInstructions = input.instructions ?? "";
|
||||
return { text: "done", toolCalls: [], rawResponseJson: JSON.stringify({ id: "resp-1" }) };
|
||||
};
|
||||
const providers = new ProviderService(connection.db, providerAdapters, testVault, undefined, fakeToolTurn);
|
||||
const providers = new ProviderService(connection.db, providerAdapters, testVault, undefined, answeringModelCheck(fakeToolTurn));
|
||||
const execution = new TaskExecutionService(connection.db, providers);
|
||||
|
||||
let orgId = "";
|
||||
@@ -189,7 +190,7 @@ describe("agent instruction wiring", () => {
|
||||
capturedPrompt = first.text;
|
||||
return { text: "done", toolCalls: [], rawResponseJson: JSON.stringify({ id: "resp-1" }) };
|
||||
};
|
||||
const providers = new ProviderService(connection.db, providerAdapters, testVault, undefined, fakeToolTurn);
|
||||
const providers = new ProviderService(connection.db, providerAdapters, testVault, undefined, answeringModelCheck(fakeToolTurn));
|
||||
const execution = new TaskExecutionService(connection.db, providers);
|
||||
let orgId = "";
|
||||
let taskId = "";
|
||||
|
||||
Reference in new issue
Block a user