docs(08.1, 09): insert agent output schema phase, update phase 9
Phase 8.1: Agent Output Schema (INSERTED) - 2 plans in 2 waves (sequential) - Defines discriminated union schema (done/question/error) - Updates ClaudeAgentManager to use --json-schema flag - Aligns MockAgentManager with new schema Phase 9: Extended Scenarios (updated) - 2 plans in 1 wave (parallel) - Now depends on Phase 8.1 - Updated scenario format references
This commit is contained in:
Lukas May
@@ -141,20 +141,32 @@ Plans:
|
||||
- [x] 08-01: Happy Path E2E Tests
|
||||
- [x] 08-02: Edge Case E2E Tests
|
||||
|
||||
#### Phase 9: Extended Scenarios & CI
|
||||
#### Phase 8.1: Agent Output Schema (INSERTED)
|
||||
|
||||
**Goal**: Additional scenario coverage + CI pipeline integration for automated test runs
|
||||
**Goal**: Define structured agent output schema (done/question/error discriminated union) and update ClaudeAgentManager to use `--json-schema` flag for validated output parsing
|
||||
**Depends on**: Phase 8
|
||||
**Research**: Unlikely (standard CI patterns)
|
||||
**Plans**: TBD
|
||||
**Research**: Unlikely (Zod schemas, Claude CLI flags documented)
|
||||
**Plans**: 2 plans
|
||||
|
||||
Plans:
|
||||
- [ ] 09-01: TBD (run /gsd:plan-phase 9 to break down)
|
||||
- [ ] 08.1-01: Agent Output Schema & ClaudeAgentManager
|
||||
- [ ] 08.1-02: MockAgentManager Schema Alignment
|
||||
|
||||
#### Phase 9: Extended Scenarios
|
||||
|
||||
**Goal**: Extended E2E scenario coverage — conflict hand-back round-trip, multi-agent parallel work, recovery/resume flows
|
||||
**Depends on**: Phase 8.1
|
||||
**Research**: Unlikely (testing existing functionality)
|
||||
**Plans**: 2 plans
|
||||
|
||||
Plans:
|
||||
- [ ] 09-01: Conflict & Parallel E2E Tests
|
||||
- [ ] 09-02: Recovery & Resume E2E Tests
|
||||
|
||||
## Progress
|
||||
|
||||
**Execution Order:**
|
||||
Phases execute in numeric order: 1 → 1.1 → 2 → 3 → 4 → 5 → 6 → 7 → 8 → 9
|
||||
Phases execute in numeric order: 1 → 1.1 → 2 → 3 → 4 → 5 → 6 → 7 → 8 → 8.1 → 9
|
||||
|
||||
| Phase | Milestone | Plans Complete | Status | Completed |
|
||||
|-------|-----------|----------------|--------|-----------|
|
||||
@@ -167,7 +179,8 @@ Phases execute in numeric order: 1 → 1.1 → 2 → 3 → 4 → 5 → 6 → 7
|
||||
| 6. Coordination | v1.0 | 3/3 | Complete | 2026-01-30 |
|
||||
| 7. Mock Agent & Test Harness | v1.1 | 2/2 | Complete | 2026-01-31 |
|
||||
| 8. E2E Scenario Tests | v1.1 | 2/2 | Complete | 2026-01-31 |
|
||||
| 9. Extended Scenarios & CI | v1.1 | 0/? | Not started | - |
|
||||
| 8.1. Agent Output Schema | v1.1 | 0/2 | Not started | - |
|
||||
| 9. Extended Scenarios | v1.1 | 0/2 | Not started | - |
|
||||
|
||||
---
|
||||
*Roadmap created: 2026-01-30*
|
||||
|
||||
269
.planning/phases/08.1-agent-output-schema/08.1-01-PLAN.md
Normal file
269
.planning/phases/08.1-agent-output-schema/08.1-01-PLAN.md
Normal file
@@ -0,0 +1,269 @@
|
||||
---
|
||||
phase: 08.1-agent-output-schema
|
||||
plan: 01
|
||||
type: execute
|
||||
wave: 1
|
||||
depends_on: []
|
||||
files_modified: [src/agent/schema.ts, src/agent/manager.ts, src/agent/types.ts]
|
||||
autonomous: true
|
||||
---
|
||||
|
||||
<objective>
|
||||
Define structured agent output schema and update ClaudeAgentManager to use `--json-schema` flag for validated output parsing.
|
||||
|
||||
Purpose: Replace broken AskUserQuestion detection with explicit agent status signaling via discriminated union schema.
|
||||
Output: Zod schema for agent output, updated ClaudeAgentManager with proper question/resume flow.
|
||||
</objective>
|
||||
|
||||
<execution_context>
|
||||
@~/.claude/get-shit-done/workflows/execute-plan.md
|
||||
@~/.claude/get-shit-done/templates/summary.md
|
||||
</execution_context>
|
||||
|
||||
<context>
|
||||
@.planning/PROJECT.md
|
||||
@.planning/ROADMAP.md
|
||||
@.planning/STATE.md
|
||||
|
||||
@src/agent/manager.ts
|
||||
@src/agent/types.ts
|
||||
@src/events/types.ts
|
||||
</context>
|
||||
|
||||
<tasks>
|
||||
|
||||
<task type="auto">
|
||||
<name>Task 1: Define agent output schema with Zod</name>
|
||||
<files>src/agent/schema.ts</files>
|
||||
<action>
|
||||
Create new file `src/agent/schema.ts` with discriminated union schema:
|
||||
|
||||
```typescript
|
||||
import { z } from 'zod';
|
||||
|
||||
// Option for questions
|
||||
const optionSchema = z.object({
|
||||
label: z.string(),
|
||||
description: z.string().optional(),
|
||||
});
|
||||
|
||||
// Discriminated union for agent output
|
||||
export const agentOutputSchema = z.discriminatedUnion('status', [
|
||||
// Agent completed successfully
|
||||
z.object({
|
||||
status: z.literal('done'),
|
||||
result: z.string(),
|
||||
filesModified: z.array(z.string()).optional(),
|
||||
}),
|
||||
|
||||
// Agent needs user input to continue
|
||||
z.object({
|
||||
status: z.literal('question'),
|
||||
question: z.string(),
|
||||
options: z.array(optionSchema).optional(),
|
||||
multiSelect: z.boolean().optional(),
|
||||
}),
|
||||
|
||||
// Agent hit unrecoverable error
|
||||
z.object({
|
||||
status: z.literal('unrecoverable_error'),
|
||||
error: z.string(),
|
||||
attempted: z.string().optional(),
|
||||
}),
|
||||
]);
|
||||
|
||||
export type AgentOutput = z.infer<typeof agentOutputSchema>;
|
||||
|
||||
// JSON Schema for --json-schema flag (convert Zod to JSON Schema)
|
||||
export const agentOutputJsonSchema = {
|
||||
type: 'object',
|
||||
oneOf: [
|
||||
{
|
||||
properties: {
|
||||
status: { const: 'done' },
|
||||
result: { type: 'string' },
|
||||
filesModified: { type: 'array', items: { type: 'string' } },
|
||||
},
|
||||
required: ['status', 'result'],
|
||||
},
|
||||
{
|
||||
properties: {
|
||||
status: { const: 'question' },
|
||||
question: { type: 'string' },
|
||||
options: {
|
||||
type: 'array',
|
||||
items: {
|
||||
type: 'object',
|
||||
properties: {
|
||||
label: { type: 'string' },
|
||||
description: { type: 'string' },
|
||||
},
|
||||
required: ['label'],
|
||||
},
|
||||
},
|
||||
multiSelect: { type: 'boolean' },
|
||||
},
|
||||
required: ['status', 'question'],
|
||||
},
|
||||
{
|
||||
properties: {
|
||||
status: { const: 'unrecoverable_error' },
|
||||
error: { type: 'string' },
|
||||
attempted: { type: 'string' },
|
||||
},
|
||||
required: ['status', 'error'],
|
||||
},
|
||||
],
|
||||
};
|
||||
```
|
||||
|
||||
Export both the Zod schema (for runtime validation) and JSON schema string (for CLI flag).
|
||||
</action>
|
||||
<verify>npx tsc --noEmit src/agent/schema.ts</verify>
|
||||
<done>Schema file created with discriminated union and JSON schema export</done>
|
||||
</task>
|
||||
|
||||
<task type="auto">
|
||||
<name>Task 2: Update ClaudeAgentManager to use schema and handle question flow</name>
|
||||
<files>src/agent/manager.ts, src/agent/types.ts</files>
|
||||
<action>
|
||||
Update ClaudeAgentManager:
|
||||
|
||||
1. **Import schema:**
|
||||
```typescript
|
||||
import { agentOutputSchema, agentOutputJsonSchema } from './schema.js';
|
||||
```
|
||||
|
||||
2. **Update spawn() to pass --json-schema:**
|
||||
```typescript
|
||||
const subprocess = execa(
|
||||
'claude',
|
||||
[
|
||||
'-p', prompt,
|
||||
'--output-format', 'json',
|
||||
'--json-schema', JSON.stringify(agentOutputJsonSchema),
|
||||
],
|
||||
{ cwd: cwd ?? worktree.path, detached: true, stdio: ['ignore', 'pipe', 'pipe'] }
|
||||
);
|
||||
```
|
||||
|
||||
3. **Update handleAgentCompletion() to parse discriminated union:**
|
||||
|
||||
Replace the current ClaudeCliResult parsing with:
|
||||
```typescript
|
||||
const cliResult = JSON.parse(stdout as string);
|
||||
|
||||
// Store session_id for resume capability
|
||||
if (cliResult.session_id) {
|
||||
await this.repository.updateSessionId(agentId, cliResult.session_id);
|
||||
}
|
||||
|
||||
// Parse the agent's structured output from result field
|
||||
const agentOutput = agentOutputSchema.parse(JSON.parse(cliResult.result));
|
||||
|
||||
switch (agentOutput.status) {
|
||||
case 'done':
|
||||
// Success path - existing logic
|
||||
active.result = { success: true, message: agentOutput.result, filesModified: agentOutput.filesModified };
|
||||
await this.repository.updateStatus(agentId, 'idle');
|
||||
// Emit agent:stopped event
|
||||
break;
|
||||
|
||||
case 'question':
|
||||
// Question path - agent needs input
|
||||
await this.repository.updateStatus(agentId, 'waiting_for_input');
|
||||
// Store question metadata for later retrieval
|
||||
active.pendingQuestion = agentOutput;
|
||||
// Emit agent:waiting event with structured question
|
||||
break;
|
||||
|
||||
case 'unrecoverable_error':
|
||||
// Error path
|
||||
active.result = { success: false, message: agentOutput.error };
|
||||
await this.repository.updateStatus(agentId, 'crashed');
|
||||
// Emit agent:crashed event
|
||||
break;
|
||||
}
|
||||
```
|
||||
|
||||
4. **Update ActiveAgent interface:**
|
||||
```typescript
|
||||
interface ActiveAgent {
|
||||
subprocess: ResultPromise;
|
||||
result?: AgentResult;
|
||||
pendingQuestion?: { question: string; options?: Array<{label: string; description?: string}>; multiSelect?: boolean };
|
||||
}
|
||||
```
|
||||
|
||||
5. **Update resume() to use same session_id:**
|
||||
- Already uses `--resume` flag with stored session_id ✓
|
||||
- Ensure prompt passed is the user's answer to the question
|
||||
|
||||
6. **Add method to get pending question:**
|
||||
```typescript
|
||||
async getPendingQuestion(agentId: string): Promise<{question: string; options?: ...} | null> {
|
||||
const active = this.activeAgents.get(agentId);
|
||||
return active?.pendingQuestion ?? null;
|
||||
}
|
||||
```
|
||||
|
||||
7. **Update AgentManager interface in types.ts:**
|
||||
Add getPendingQuestion method signature.
|
||||
|
||||
8. **Remove the hacky string matching** in handleAgentError() for 'waiting for input' detection - no longer needed.
|
||||
</action>
|
||||
<verify>npm run build && npm run typecheck</verify>
|
||||
<done>ClaudeAgentManager uses --json-schema, parses discriminated union, handles question/resume flow correctly</done>
|
||||
</task>
|
||||
|
||||
<task type="auto">
|
||||
<name>Task 3: Update AgentWaitingEvent to include structured question data</name>
|
||||
<files>src/events/types.ts</files>
|
||||
<action>
|
||||
Update AgentWaitingEvent payload to include structured question data:
|
||||
|
||||
```typescript
|
||||
export interface AgentWaitingEvent extends BaseEvent {
|
||||
type: 'agent:waiting';
|
||||
payload: {
|
||||
agentId: string;
|
||||
name: string;
|
||||
taskId: string;
|
||||
sessionId: string;
|
||||
question: string;
|
||||
options?: Array<{ label: string; description?: string }>;
|
||||
multiSelect?: boolean;
|
||||
};
|
||||
}
|
||||
```
|
||||
|
||||
This allows event consumers to receive the full question structure, not just a plain string.
|
||||
</action>
|
||||
<verify>npm run typecheck</verify>
|
||||
<done>AgentWaitingEvent includes structured question metadata</done>
|
||||
</task>
|
||||
|
||||
</tasks>
|
||||
|
||||
<verification>
|
||||
Before declaring plan complete:
|
||||
- [ ] `npm run build` succeeds
|
||||
- [ ] `npm run typecheck` passes
|
||||
- [ ] Schema exports both Zod and JSON Schema formats
|
||||
- [ ] ClaudeAgentManager passes --json-schema to CLI
|
||||
- [ ] Question status triggers waiting_for_input with stored question
|
||||
- [ ] Resume uses same session_id
|
||||
</verification>
|
||||
|
||||
<success_criteria>
|
||||
|
||||
- All tasks completed
|
||||
- All verification checks pass
|
||||
- Agent output is validated against schema
|
||||
- Question -> answer -> resume flow uses same session_id
|
||||
- No breaking changes to existing tests (MockAgentManager updated separately)
|
||||
</success_criteria>
|
||||
|
||||
<output>
|
||||
After completion, create `.planning/phases/08.1-agent-output-schema/08.1-01-SUMMARY.md`
|
||||
</output>
|
||||
272
.planning/phases/08.1-agent-output-schema/08.1-02-PLAN.md
Normal file
272
.planning/phases/08.1-agent-output-schema/08.1-02-PLAN.md
Normal file
@@ -0,0 +1,272 @@
|
||||
---
|
||||
phase: 08.1-agent-output-schema
|
||||
plan: 02
|
||||
type: execute
|
||||
wave: 2
|
||||
depends_on: ["08.1-01"]
|
||||
files_modified: [src/agent/mock-manager.ts, src/test/harness.ts, src/agent/mock-manager.test.ts]
|
||||
autonomous: true
|
||||
---
|
||||
|
||||
<objective>
|
||||
Update MockAgentManager to use the new agent output schema, enabling proper testing of question/resume flows.
|
||||
|
||||
Purpose: Align mock implementation with real ClaudeAgentManager behavior for accurate E2E testing.
|
||||
Output: MockAgentManager that simulates structured output schema with question/resume support.
|
||||
</objective>
|
||||
|
||||
<execution_context>
|
||||
@~/.claude/get-shit-done/workflows/execute-plan.md
|
||||
@~/.claude/get-shit-done/templates/summary.md
|
||||
</execution_context>
|
||||
|
||||
<context>
|
||||
@.planning/PROJECT.md
|
||||
@.planning/ROADMAP.md
|
||||
@.planning/STATE.md
|
||||
@.planning/phases/08.1-agent-output-schema/08.1-01-SUMMARY.md
|
||||
|
||||
@src/agent/schema.ts
|
||||
@src/agent/mock-manager.ts
|
||||
@src/agent/types.ts
|
||||
@src/test/harness.ts
|
||||
</context>
|
||||
|
||||
<tasks>
|
||||
|
||||
<task type="auto">
|
||||
<name>Task 1: Update MockAgentManager to use schema-aligned scenarios</name>
|
||||
<files>src/agent/mock-manager.ts</files>
|
||||
<action>
|
||||
Update MockAgentManager to match the new schema behavior:
|
||||
|
||||
1. **Update MockAgentScenario to align with schema:**
|
||||
```typescript
|
||||
export type MockAgentScenario =
|
||||
| { status: 'done'; result?: string; filesModified?: string[]; delay?: number }
|
||||
| { status: 'question'; question: string; options?: Array<{label: string; description?: string}>; multiSelect?: boolean; delay?: number }
|
||||
| { status: 'unrecoverable_error'; error: string; attempted?: string; delay?: number };
|
||||
```
|
||||
|
||||
2. **Update DEFAULT_SCENARIO:**
|
||||
```typescript
|
||||
const DEFAULT_SCENARIO: MockAgentScenario = {
|
||||
status: 'done',
|
||||
result: 'Task completed successfully',
|
||||
filesModified: [],
|
||||
delay: 0,
|
||||
};
|
||||
```
|
||||
|
||||
3. **Update MockAgentRecord to store pending question:**
|
||||
```typescript
|
||||
interface MockAgentRecord {
|
||||
info: AgentInfo;
|
||||
scenario: MockAgentScenario;
|
||||
result?: AgentResult;
|
||||
pendingQuestion?: { question: string; options?: Array<{label: string; description?: string}>; multiSelect?: boolean };
|
||||
completionTimer?: ReturnType<typeof setTimeout>;
|
||||
}
|
||||
```
|
||||
|
||||
4. **Update completeAgent() to handle new schema:**
|
||||
```typescript
|
||||
private completeAgent(agentId: string, scenario: MockAgentScenario): void {
|
||||
const record = this.agents.get(agentId);
|
||||
if (!record) return;
|
||||
|
||||
switch (scenario.status) {
|
||||
case 'done':
|
||||
record.result = {
|
||||
success: true,
|
||||
message: scenario.result ?? 'Task completed successfully',
|
||||
filesModified: scenario.filesModified,
|
||||
};
|
||||
record.info.status = 'idle';
|
||||
// Emit agent:stopped event
|
||||
break;
|
||||
|
||||
case 'question':
|
||||
record.info.status = 'waiting_for_input';
|
||||
record.pendingQuestion = {
|
||||
question: scenario.question,
|
||||
options: scenario.options,
|
||||
multiSelect: scenario.multiSelect,
|
||||
};
|
||||
// Emit agent:waiting event with full question data
|
||||
if (this.eventBus) {
|
||||
const event: AgentWaitingEvent = {
|
||||
type: 'agent:waiting',
|
||||
timestamp: new Date(),
|
||||
payload: {
|
||||
agentId,
|
||||
name: record.info.name,
|
||||
taskId: record.info.taskId,
|
||||
sessionId: record.info.sessionId ?? '',
|
||||
question: scenario.question,
|
||||
options: scenario.options,
|
||||
multiSelect: scenario.multiSelect,
|
||||
},
|
||||
};
|
||||
this.eventBus.emit(event);
|
||||
}
|
||||
break;
|
||||
|
||||
case 'unrecoverable_error':
|
||||
record.result = {
|
||||
success: false,
|
||||
message: scenario.error,
|
||||
};
|
||||
record.info.status = 'crashed';
|
||||
// Emit agent:crashed event
|
||||
break;
|
||||
}
|
||||
}
|
||||
```
|
||||
|
||||
5. **Add getPendingQuestion() method:**
|
||||
```typescript
|
||||
async getPendingQuestion(agentId: string): Promise<{question: string; options?: Array<{label: string; description?: string}>; multiSelect?: boolean} | null> {
|
||||
const record = this.agents.get(agentId);
|
||||
return record?.pendingQuestion ?? null;
|
||||
}
|
||||
```
|
||||
|
||||
6. **Update resume() to clear pending question:**
|
||||
After successful resume, clear `record.pendingQuestion = undefined`.
|
||||
</action>
|
||||
<verify>npm run typecheck</verify>
|
||||
<done>MockAgentManager uses schema-aligned scenarios with structured question support</done>
|
||||
</task>
|
||||
|
||||
<task type="auto">
|
||||
<name>Task 2: Update TestHarness setAgentScenario helper</name>
|
||||
<files>src/test/harness.ts</files>
|
||||
<action>
|
||||
Update TestHarness to use new scenario format:
|
||||
|
||||
1. **Update setAgentScenario type:**
|
||||
```typescript
|
||||
import type { MockAgentScenario } from '../agent/mock-manager.js';
|
||||
|
||||
// In TestHarness interface:
|
||||
setAgentScenario(agentName: string, scenario: MockAgentScenario): void;
|
||||
```
|
||||
|
||||
2. **Add helper for common scenarios:**
|
||||
```typescript
|
||||
// In TestHarness interface, add convenience methods:
|
||||
setAgentDone(agentName: string, result?: string): void;
|
||||
setAgentQuestion(agentName: string, question: string, options?: Array<{label: string; description?: string}>): void;
|
||||
setAgentError(agentName: string, error: string): void;
|
||||
|
||||
// Implementation:
|
||||
setAgentDone: (name, result) => agentManager.setScenario(name, { status: 'done', result }),
|
||||
setAgentQuestion: (name, question, options) => agentManager.setScenario(name, { status: 'question', question, options }),
|
||||
setAgentError: (name, error) => agentManager.setScenario(name, { status: 'unrecoverable_error', error }),
|
||||
```
|
||||
|
||||
3. **Add getPendingQuestion to harness:**
|
||||
```typescript
|
||||
getPendingQuestion(agentId: string): Promise<{question: string; options?: ...} | null>;
|
||||
```
|
||||
</action>
|
||||
<verify>npm run typecheck</verify>
|
||||
<done>TestHarness updated with new scenario helpers</done>
|
||||
</task>
|
||||
|
||||
<task type="auto">
|
||||
<name>Task 3: Update MockAgentManager tests</name>
|
||||
<files>src/agent/mock-manager.test.ts</files>
|
||||
<action>
|
||||
Update existing MockAgentManager tests to use new scenario format:
|
||||
|
||||
1. **Update scenario declarations:**
|
||||
Change from:
|
||||
```typescript
|
||||
{ outcome: 'success', message: '...' }
|
||||
{ outcome: 'crash', message: '...' }
|
||||
{ outcome: 'waiting_for_input', question: '...' }
|
||||
```
|
||||
To:
|
||||
```typescript
|
||||
{ status: 'done', result: '...' }
|
||||
{ status: 'unrecoverable_error', error: '...' }
|
||||
{ status: 'question', question: '...', options: [...] }
|
||||
```
|
||||
|
||||
2. **Add test for structured question data:**
|
||||
```typescript
|
||||
it('emits agent:waiting with structured question data', async () => {
|
||||
vi.useFakeTimers();
|
||||
manager.setScenario('test-agent', {
|
||||
status: 'question',
|
||||
question: 'Which database?',
|
||||
options: [
|
||||
{ label: 'PostgreSQL', description: 'Full-featured' },
|
||||
{ label: 'SQLite', description: 'Lightweight' },
|
||||
],
|
||||
multiSelect: false,
|
||||
});
|
||||
|
||||
await manager.spawn({ name: 'test-agent', taskId: 'task-1', prompt: 'test' });
|
||||
await vi.runAllTimersAsync();
|
||||
|
||||
const events = eventBus.getEventsByType('agent:waiting');
|
||||
expect(events[0].payload.options).toHaveLength(2);
|
||||
expect(events[0].payload.options[0].label).toBe('PostgreSQL');
|
||||
});
|
||||
```
|
||||
|
||||
3. **Add test for getPendingQuestion:**
|
||||
```typescript
|
||||
it('stores pending question for retrieval', async () => {
|
||||
vi.useFakeTimers();
|
||||
manager.setScenario('test-agent', {
|
||||
status: 'question',
|
||||
question: 'Which database?',
|
||||
options: [{ label: 'PostgreSQL' }],
|
||||
});
|
||||
|
||||
const agent = await manager.spawn({ name: 'test-agent', taskId: 'task-1', prompt: 'test' });
|
||||
await vi.runAllTimersAsync();
|
||||
|
||||
const pending = await manager.getPendingQuestion(agent.id);
|
||||
expect(pending?.question).toBe('Which database?');
|
||||
expect(pending?.options).toHaveLength(1);
|
||||
});
|
||||
```
|
||||
|
||||
4. **Add test for resume clearing pending question:**
|
||||
```typescript
|
||||
it('clears pending question after resume', async () => {
|
||||
// Setup question scenario, resume, verify pendingQuestion is null
|
||||
});
|
||||
```
|
||||
</action>
|
||||
<verify>npm test src/agent/mock-manager.test.ts -- --run</verify>
|
||||
<done>MockAgentManager tests updated and passing with new schema</done>
|
||||
</task>
|
||||
|
||||
</tasks>
|
||||
|
||||
<verification>
|
||||
Before declaring plan complete:
|
||||
- [ ] `npm run build` succeeds
|
||||
- [ ] `npm run typecheck` passes
|
||||
- [ ] `npm test src/agent/mock-manager.test.ts -- --run` passes
|
||||
- [ ] Existing E2E tests still pass (may need scenario format updates)
|
||||
</verification>
|
||||
|
||||
<success_criteria>
|
||||
|
||||
- All tasks completed
|
||||
- All verification checks pass
|
||||
- MockAgentManager matches ClaudeAgentManager behavior
|
||||
- TestHarness provides convenient scenario helpers
|
||||
- All mock-manager tests pass with new schema
|
||||
</success_criteria>
|
||||
|
||||
<output>
|
||||
After completion, create `.planning/phases/08.1-agent-output-schema/08.1-02-SUMMARY.md`
|
||||
</output>
|
||||
@@ -6,6 +6,7 @@ wave: 1
|
||||
depends_on: []
|
||||
files_modified: [src/test/e2e/extended-scenarios.test.ts]
|
||||
autonomous: true
|
||||
phase_depends_on: [08.1-agent-output-schema]
|
||||
---
|
||||
|
||||
<objective>
|
||||
@@ -27,6 +28,7 @@ Output: Extended E2E test file with conflict round-trip and parallel completion
|
||||
@.planning/phases/09-extended-scenarios/09-CONTEXT.md
|
||||
@.planning/phases/08-e2e-scenario-tests/08-01-SUMMARY.md
|
||||
@.planning/phases/08-e2e-scenario-tests/08-02-SUMMARY.md
|
||||
@.planning/phases/08.1-agent-output-schema/08.1-02-SUMMARY.md
|
||||
|
||||
@src/test/harness.ts
|
||||
@src/test/fixtures.ts
|
||||
@@ -71,7 +73,7 @@ Create new test file with describe block "Conflict hand-back round-trip". Test t
|
||||
Use same patterns as edge-cases.test.ts:
|
||||
- vi.useFakeTimers() for async control
|
||||
- Pre-seed idle agent before dispatch
|
||||
- harness.setAgentScenario for agent behavior
|
||||
- harness.setAgentScenario with new schema format: `{ status: 'done' | 'question' | 'unrecoverable_error', ... }`
|
||||
- harness.worktreeManager.setMergeResult for conflict injection
|
||||
- Manual agentRepository.create for coordination tests
|
||||
</action>
|
||||
|
||||
@@ -6,6 +6,7 @@ wave: 1
|
||||
depends_on: []
|
||||
files_modified: [src/test/e2e/recovery-scenarios.test.ts]
|
||||
autonomous: true
|
||||
phase_depends_on: [08.1-agent-output-schema]
|
||||
---
|
||||
|
||||
<objective>
|
||||
@@ -26,6 +27,7 @@ Output: Recovery scenarios test file with state persistence and Q&A flow tests.
|
||||
@.planning/STATE.md
|
||||
@.planning/phases/09-extended-scenarios/09-CONTEXT.md
|
||||
@.planning/phases/08-e2e-scenario-tests/08-02-SUMMARY.md
|
||||
@.planning/phases/08.1-agent-output-schema/08.1-02-SUMMARY.md
|
||||
|
||||
@src/test/harness.ts
|
||||
@src/test/fixtures.ts
|
||||
@@ -117,7 +119,11 @@ Add describe block "Agent Q&A extended scenarios" to the test file. Test scenari
|
||||
- Now complete task
|
||||
- Verify: proper state transitions
|
||||
|
||||
Use edge-cases.test.ts patterns for waiting/resume flow.
|
||||
Use new schema format for scenarios:
|
||||
- `{ status: 'question', question: '...', options: [...] }` for questions
|
||||
- `{ status: 'done', result: '...' }` for success
|
||||
- `{ status: 'unrecoverable_error', error: '...' }` for failures
|
||||
- harness.getPendingQuestion() to retrieve structured question data
|
||||
</action>
|
||||
<verify>npm test src/test/e2e/recovery-scenarios.test.ts -- --run passes</verify>
|
||||
<done>4 Q&A tests passing, proving extended question flows work</done>
|
||||
|
||||
Reference in New Issue
Block a user