chore(07-02): bring in plan 01 prerequisites
Cherry-pick plan 01 task quality gate validation module and its integration into IPC, API server, and automation service. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
1 parent
c1de05a7f1
commit
4d1085fc04
6 files changed
+412
-2
No files matched your search
@@ -8,6 +8,7 @@ import { canAssignTask, canControlAgent, canManageGoal, canManageOrganization, c
|
||||
import { resolveApprovalDecision, shouldAutoApproveApproval } from "./approval-policy";
|
||||
import type { AppDatabase } from "./database";
|
||||
import { inferTaskType } from "./connector-matching";
|
||||
import { validateTaskCreation } from "./task-quality-gates";
|
||||
import { dispatchAgentMessage } from "./message-dispatch";
|
||||
import type { DesktopEvent, SectionId } from "@shared/types";
|
||||
|
||||
@@ -561,6 +562,12 @@ export class AgentApiServer {
|
||||
this.json(res, 403, { error: "Current agent is not allowed to assign this task to the selected agent" });
|
||||
return;
|
||||
}
|
||||
const isHeartbeat = input.title.startsWith("[Heartbeat]");
|
||||
const taskValidation = validateTaskCreation({ title: input.title, description: input.description ?? "" }, isHeartbeat);
|
||||
if (!taskValidation.valid) {
|
||||
this.json(res, 422, { error: taskValidation.errors.join("; ") });
|
||||
return;
|
||||
}
|
||||
|
||||
let id: string;
|
||||
try {
|
||||
@@ -1582,6 +1589,14 @@ export class AgentApiServer {
|
||||
if (actionItems.length > 0) {
|
||||
for (const item of actionItems) {
|
||||
if (item.title && typeof item.title === "string") {
|
||||
const meetingTaskValidation = validateTaskCreation(
|
||||
{ title: item.title, description: item.description ?? `Action item from meeting: ${data.title ?? existing.title}` },
|
||||
false,
|
||||
);
|
||||
if (!meetingTaskValidation.valid) {
|
||||
// Log but skip invalid meeting action items rather than failing the whole request
|
||||
continue;
|
||||
}
|
||||
const taskId = this.db.saveTask({
|
||||
companyId: token.companyId,
|
||||
title: item.title,
|
||||
|
||||
@@ -8,6 +8,7 @@ import type {
|
||||
RunStatus,
|
||||
} from "@shared/types";
|
||||
import { dispatchAgentMessage } from "./message-dispatch";
|
||||
import { validateTaskCreation } from "./task-quality-gates";
|
||||
import type { NotificationOptions } from "./notification-service";
|
||||
|
||||
/** Document types considered important enough to gate behind user review when reviewDeliverables is enabled. */
|
||||
@@ -268,6 +269,12 @@ export function createAutomationService({
|
||||
const description = (config.description as string) ?? "";
|
||||
const priority = (config.priority as string) ?? "medium";
|
||||
const assigneeAgentId = (config.assigneeAgentId as string) ?? null;
|
||||
const isHeartbeat = title.startsWith("[Heartbeat]");
|
||||
const taskValidation = validateTaskCreation({ title, description }, isHeartbeat);
|
||||
if (!taskValidation.valid) {
|
||||
logger.warn(`[automation] Skipping create_task — validation failed: ${taskValidation.errors.join("; ")}`);
|
||||
break;
|
||||
}
|
||||
const taskId = db.saveTask({
|
||||
companyId: ctx.companyId,
|
||||
projectId: ctx.projectId ?? (config.projectId as string) ?? null,
|
||||
|
||||
@@ -11,6 +11,7 @@ import {
|
||||
import { getConnectorExecutionReadinessIssue } from "@shared/connector-policy";
|
||||
import { resolveApprovalDecision, shouldAutoApproveApproval } from "./approval-policy";
|
||||
import { fail, ok, type RegisterCoreHandlersDependencies } from "./ipc-core-common";
|
||||
import { validateTaskCreation, checkCapabilityMatch } from "./task-quality-gates";
|
||||
|
||||
export function registerCoreWorkGraphHandlers({
|
||||
registerHandle,
|
||||
@@ -314,6 +315,13 @@ export function registerCoreWorkGraphHandlers({
|
||||
return fail("INVALID_COMPANY_REFERENCE", `The selected ${invalidReference.label} does not belong to the current company.`);
|
||||
}
|
||||
const isNew = !parsed.id;
|
||||
if (isNew) {
|
||||
const isHeartbeat = parsed.title.startsWith("[Heartbeat]");
|
||||
const validation = validateTaskCreation(parsed, isHeartbeat);
|
||||
if (!validation.valid) {
|
||||
return fail("TASK_VALIDATION_FAILED", validation.errors.join("; "));
|
||||
}
|
||||
}
|
||||
let previousStatus: string | null = null;
|
||||
let previousAssigneeAgentId: string | null = null;
|
||||
if (!isNew) {
|
||||
@@ -355,8 +363,17 @@ export function registerCoreWorkGraphHandlers({
|
||||
wakeAgentIfPossible(parsed.assigneeAgentId, parsed.companyId, "assignment");
|
||||
}
|
||||
}
|
||||
let capabilityWarning: string | undefined;
|
||||
if (isNew && parsed.assigneeAgentId) {
|
||||
try {
|
||||
const match = checkCapabilityMatch(db, parsed.assigneeAgentId, parsed.description, parsed.companyId);
|
||||
if (!match.matched) {
|
||||
capabilityWarning = match.warning;
|
||||
}
|
||||
} catch { /* capability check is advisory — never block task creation */ }
|
||||
}
|
||||
publishDomainChanged();
|
||||
return ok(id);
|
||||
return ok({ id, capabilityWarning });
|
||||
});
|
||||
|
||||
registerHandle(IPC_CHANNELS.deleteAgent, async (_event, payload) => {
|
||||
|
||||
@@ -0,0 +1,163 @@
|
||||
/**
|
||||
* Task Quality Gates — pure validation functions for task creation.
|
||||
*
|
||||
* Called by IPC handlers, API server, and automation service to enforce
|
||||
* minimum description quality, acceptance criteria, and capability matching.
|
||||
*/
|
||||
|
||||
// ---------------------------------------------------------------------------
|
||||
// Types
|
||||
// ---------------------------------------------------------------------------
|
||||
|
||||
export interface TaskValidationResult {
|
||||
valid: boolean;
|
||||
errors: string[];
|
||||
warnings: string[];
|
||||
}
|
||||
|
||||
export interface CapabilityMatchResult {
|
||||
matched: boolean;
|
||||
warning?: string;
|
||||
}
|
||||
|
||||
/** Minimal DB interface for capability matching — keeps the function testable with mocks.
|
||||
* Skill methods are optional since the skill system may not be implemented yet (graceful degradation per D-07). */
|
||||
export interface CapabilityMatchDb {
|
||||
getAgent(id: string): { name: string; capabilities: string };
|
||||
listAgentSkillAssignments?(id: string): string[];
|
||||
getCompanySkill?(companyId: string, skillId: string): { name: string; description: string | null; slug: string } | null;
|
||||
}
|
||||
|
||||
// ---------------------------------------------------------------------------
|
||||
// Constants
|
||||
// ---------------------------------------------------------------------------
|
||||
|
||||
const MIN_DESCRIPTION_LENGTH = 50;
|
||||
|
||||
export const ACCEPTANCE_CRITERIA_KEYWORDS: readonly string[] = [
|
||||
"acceptance criteria",
|
||||
"done when",
|
||||
"success criteria",
|
||||
"definition of done",
|
||||
"requirements:",
|
||||
"must:",
|
||||
"should:",
|
||||
"- [ ]",
|
||||
"- [x]",
|
||||
"deliverable:",
|
||||
"expected outcome",
|
||||
"exit criteria",
|
||||
];
|
||||
|
||||
// ---------------------------------------------------------------------------
|
||||
// checkAcceptanceCriteria
|
||||
// ---------------------------------------------------------------------------
|
||||
|
||||
export function checkAcceptanceCriteria(description: string): { hasCriteria: boolean; suggestion: string } {
|
||||
const lower = description.toLowerCase();
|
||||
const found = ACCEPTANCE_CRITERIA_KEYWORDS.some((kw) => lower.includes(kw));
|
||||
|
||||
if (found) {
|
||||
return { hasCriteria: true, suggestion: "" };
|
||||
}
|
||||
|
||||
return {
|
||||
hasCriteria: false,
|
||||
suggestion:
|
||||
'Task description should include acceptance criteria. Add a "Done when:" or "Acceptance criteria:" section, or use checkbox items (- [ ]) to define success conditions.',
|
||||
};
|
||||
}
|
||||
|
||||
// ---------------------------------------------------------------------------
|
||||
// validateTaskCreation
|
||||
// ---------------------------------------------------------------------------
|
||||
|
||||
export function validateTaskCreation(
|
||||
input: { title: string; description: string },
|
||||
isHeartbeat: boolean,
|
||||
): TaskValidationResult {
|
||||
// Heartbeat tasks bypass all validation
|
||||
if (isHeartbeat) {
|
||||
return { valid: true, errors: [], warnings: [] };
|
||||
}
|
||||
|
||||
const errors: string[] = [];
|
||||
const warnings: string[] = [];
|
||||
|
||||
// TASK-01: Minimum description length
|
||||
if (input.description.length < MIN_DESCRIPTION_LENGTH) {
|
||||
errors.push(
|
||||
`Task description must be at least ${MIN_DESCRIPTION_LENGTH} characters (currently ${input.description.length}). Add details about what needs to be done, expected behavior, and any constraints.`,
|
||||
);
|
||||
}
|
||||
|
||||
// TASK-02: Acceptance criteria detection (only check if length passes)
|
||||
if (input.description.length >= MIN_DESCRIPTION_LENGTH) {
|
||||
const criteria = checkAcceptanceCriteria(input.description);
|
||||
if (!criteria.hasCriteria) {
|
||||
errors.push(criteria.suggestion);
|
||||
}
|
||||
}
|
||||
|
||||
return { valid: errors.length === 0, errors, warnings };
|
||||
}
|
||||
|
||||
// ---------------------------------------------------------------------------
|
||||
// checkCapabilityMatch
|
||||
// ---------------------------------------------------------------------------
|
||||
|
||||
/** Tokenize text into words >3 chars, same pattern as database.ts autoAssignTask */
|
||||
function tokenize(text: string): string[] {
|
||||
return text.toLowerCase().split(/\W+/).filter((w) => w.length > 3);
|
||||
}
|
||||
|
||||
export function checkCapabilityMatch(
|
||||
db: CapabilityMatchDb,
|
||||
agentId: string,
|
||||
taskDescription: string,
|
||||
companyId: string,
|
||||
): CapabilityMatchResult {
|
||||
const taskWords = tokenize(taskDescription);
|
||||
|
||||
// Nothing to match against — task description too short/vague
|
||||
if (taskWords.length === 0) {
|
||||
return { matched: true };
|
||||
}
|
||||
|
||||
const agent = db.getAgent(agentId);
|
||||
const capWords = tokenize(agent.capabilities);
|
||||
const capabilityOverlap = taskWords.filter((w) => capWords.includes(w)).length;
|
||||
|
||||
// Check skill assignments for additional matching (graceful degradation)
|
||||
let skillOverlap = 0;
|
||||
try {
|
||||
if (db.listAgentSkillAssignments && db.getCompanySkill) {
|
||||
const skillIds = db.listAgentSkillAssignments(agentId);
|
||||
for (const skillId of skillIds) {
|
||||
const skill = db.getCompanySkill(companyId, skillId);
|
||||
if (skill) {
|
||||
const skillWords = tokenize(`${skill.name} ${skill.description ?? ""} ${skill.slug}`);
|
||||
skillOverlap += taskWords.filter((w) => skillWords.includes(w)).length;
|
||||
}
|
||||
}
|
||||
}
|
||||
} catch {
|
||||
// Skill system may not be available — graceful degradation per D-07
|
||||
}
|
||||
|
||||
// Direct capability or skill overlap found
|
||||
if (capabilityOverlap > 0 || skillOverlap > 0) {
|
||||
return { matched: true };
|
||||
}
|
||||
|
||||
// No capabilities defined at all — graceful degradation (Pitfall 4)
|
||||
if (capWords.length === 0 && skillOverlap === 0) {
|
||||
return { matched: true };
|
||||
}
|
||||
|
||||
// Agent has capabilities but no overlap — warn
|
||||
return {
|
||||
matched: false,
|
||||
warning: `Agent "${agent.name}" may lack skills for this task. Their capabilities (${agent.capabilities.slice(0, 80)}...) have no overlap with the task description. Consider reassigning.`,
|
||||
};
|
||||
}
|
||||
@@ -518,6 +518,11 @@ export type DesktopResult<T> =
|
||||
| { ok: true; data: T }
|
||||
| { ok: false; error: { code: string; message: string } };
|
||||
|
||||
export interface SaveTaskResult {
|
||||
id: string;
|
||||
capabilityWarning?: string;
|
||||
}
|
||||
|
||||
export type DesktopApi = {
|
||||
loadSnapshot: () => Promise<DesktopResult<ProfileSnapshot>>;
|
||||
pickDirectory: () => Promise<DesktopResult<string | null>>;
|
||||
@@ -531,7 +536,7 @@ export type DesktopApi = {
|
||||
requestHire: (input: z.infer<typeof hireRequestSchema>) => Promise<DesktopResult<{ agentId: string; approvalId: string; autoApproved: boolean }>>;
|
||||
saveGoal: (input: z.infer<typeof goalInputSchema>) => Promise<DesktopResult<string>>;
|
||||
saveProject: (input: z.infer<typeof projectInputSchema>) => Promise<DesktopResult<string>>;
|
||||
saveTask: (input: z.infer<typeof taskInputSchema>) => Promise<DesktopResult<string>>;
|
||||
saveTask: (input: z.infer<typeof taskInputSchema>) => Promise<DesktopResult<SaveTaskResult>>;
|
||||
requestApproval: (input: z.infer<typeof approvalInputSchema>) => Promise<DesktopResult<string>>;
|
||||
decideApproval: (input: z.infer<typeof approvalDecisionSchema>) => Promise<DesktopResult<boolean>>;
|
||||
saveSecret: (input: z.infer<typeof secretInputSchema>) => Promise<DesktopResult<string>>;
|
||||
|
||||
@@ -0,0 +1,203 @@
|
||||
import { describe, expect, it } from "vitest";
|
||||
import {
|
||||
validateTaskCreation,
|
||||
checkAcceptanceCriteria,
|
||||
checkCapabilityMatch,
|
||||
ACCEPTANCE_CRITERIA_KEYWORDS,
|
||||
} from "../../src/main/task-quality-gates";
|
||||
import type { TaskValidationResult } from "../../src/main/task-quality-gates";
|
||||
|
||||
// --- TASK-01: validateTaskCreation tests ---
|
||||
|
||||
describe("validateTaskCreation", () => {
|
||||
it("rejects description with 5 chars when not heartbeat", () => {
|
||||
const result = validateTaskCreation({ title: "Fix bug", description: "Short" }, false);
|
||||
expect(result.valid).toBe(false);
|
||||
expect(result.errors.some(e => e.includes("at least 50 characters"))).toBe(true);
|
||||
});
|
||||
|
||||
it("rejects description with exactly 49 chars when not heartbeat", () => {
|
||||
const desc = "A".repeat(49);
|
||||
const result = validateTaskCreation({ title: "Fix bug", description: desc }, false);
|
||||
expect(result.valid).toBe(false);
|
||||
expect(result.errors.some(e => e.includes("at least 50 characters"))).toBe(true);
|
||||
});
|
||||
|
||||
it("accepts description with exactly 50 chars and acceptance keyword when not heartbeat", () => {
|
||||
const desc = "Done when: the login page renders correctly." + "......";
|
||||
// Ensure exactly 50 chars
|
||||
const padded = desc.slice(0, 50).padEnd(50, ".");
|
||||
expect(padded.length).toBe(50);
|
||||
// Must contain "done when"
|
||||
const result = validateTaskCreation({ title: "Build login page", description: padded }, false);
|
||||
expect(result.valid).toBe(true);
|
||||
expect(result.errors).toHaveLength(0);
|
||||
});
|
||||
|
||||
it("always returns valid for heartbeat tasks regardless of description", () => {
|
||||
const result = validateTaskCreation({ title: "[Heartbeat] Check status", description: "Short" }, true);
|
||||
expect(result.valid).toBe(true);
|
||||
expect(result.errors).toHaveLength(0);
|
||||
expect(result.warnings).toHaveLength(0);
|
||||
});
|
||||
|
||||
it("rejects empty description when not heartbeat", () => {
|
||||
const result = validateTaskCreation({ title: "Some task", description: "" }, false);
|
||||
expect(result.valid).toBe(false);
|
||||
expect(result.errors.some(e => e.includes("at least 50 characters"))).toBe(true);
|
||||
});
|
||||
|
||||
it("rejects 60-char description without acceptance criteria keywords", () => {
|
||||
const desc = "This is a description that is long enough to pass length check";
|
||||
expect(desc.length).toBeGreaterThanOrEqual(50);
|
||||
const result = validateTaskCreation({ title: "A task", description: desc }, false);
|
||||
expect(result.valid).toBe(false);
|
||||
expect(result.errors.some(e => e.toLowerCase().includes("acceptance criteria"))).toBe(true);
|
||||
});
|
||||
|
||||
it("returns valid=true when description has both length and acceptance criteria", () => {
|
||||
const desc = "Implement the authentication flow for the login page. Done when: all tests pass and the form submits.";
|
||||
expect(desc.length).toBeGreaterThanOrEqual(50);
|
||||
const result = validateTaskCreation({ title: "Auth flow", description: desc }, false);
|
||||
expect(result.valid).toBe(true);
|
||||
expect(result.errors).toHaveLength(0);
|
||||
});
|
||||
|
||||
it("includes current character count in length error message", () => {
|
||||
const result = validateTaskCreation({ title: "Fix", description: "Short" }, false);
|
||||
expect(result.errors.some(e => e.includes("currently 5"))).toBe(true);
|
||||
});
|
||||
});
|
||||
|
||||
// --- TASK-02: checkAcceptanceCriteria tests ---
|
||||
|
||||
describe("checkAcceptanceCriteria", () => {
|
||||
it("detects 'done when' keyword", () => {
|
||||
const result = checkAcceptanceCriteria("This task is done when all tests pass and deployment succeeds");
|
||||
expect(result.hasCriteria).toBe(true);
|
||||
});
|
||||
|
||||
it("detects 'Acceptance Criteria:' case insensitive", () => {
|
||||
const result = checkAcceptanceCriteria("Acceptance Criteria: the form validates all fields correctly");
|
||||
expect(result.hasCriteria).toBe(true);
|
||||
});
|
||||
|
||||
it("detects '- [ ] first item' checkbox", () => {
|
||||
const result = checkAcceptanceCriteria("Tasks to complete:\n- [ ] first item\n- [ ] second item");
|
||||
expect(result.hasCriteria).toBe(true);
|
||||
});
|
||||
|
||||
it("detects '- [x] completed item' checked checkbox", () => {
|
||||
const result = checkAcceptanceCriteria("Progress:\n- [x] completed item\n- [ ] pending item");
|
||||
expect(result.hasCriteria).toBe(true);
|
||||
});
|
||||
|
||||
it("detects 'must:' keyword", () => {
|
||||
const result = checkAcceptanceCriteria("The implementation must: handle all edge cases gracefully");
|
||||
expect(result.hasCriteria).toBe(true);
|
||||
});
|
||||
|
||||
it("detects 'requirements:' keyword", () => {
|
||||
const result = checkAcceptanceCriteria("Technical requirements: use React 19 with TypeScript strict");
|
||||
expect(result.hasCriteria).toBe(true);
|
||||
});
|
||||
|
||||
it("detects 'success criteria' keyword", () => {
|
||||
const result = checkAcceptanceCriteria("Success criteria for this task include passing all tests");
|
||||
expect(result.hasCriteria).toBe(true);
|
||||
});
|
||||
|
||||
it("detects 'deliverable:' keyword", () => {
|
||||
const result = checkAcceptanceCriteria("Primary deliverable: a working API endpoint with docs");
|
||||
expect(result.hasCriteria).toBe(true);
|
||||
});
|
||||
|
||||
it("detects 'expected outcome' keyword", () => {
|
||||
const result = checkAcceptanceCriteria("Expected outcome is a fully tested authentication module");
|
||||
expect(result.hasCriteria).toBe(true);
|
||||
});
|
||||
|
||||
it("detects 'exit criteria' keyword", () => {
|
||||
const result = checkAcceptanceCriteria("Exit criteria for this sprint includes shipped feature");
|
||||
expect(result.hasCriteria).toBe(true);
|
||||
});
|
||||
|
||||
it("returns hasCriteria=false with suggestion for description with no keywords", () => {
|
||||
const result = checkAcceptanceCriteria("Build a login page with email and password fields");
|
||||
expect(result.hasCriteria).toBe(false);
|
||||
expect(result.suggestion).toContain("Done when:");
|
||||
});
|
||||
});
|
||||
|
||||
// --- Verify ACCEPTANCE_CRITERIA_KEYWORDS constant ---
|
||||
|
||||
describe("ACCEPTANCE_CRITERIA_KEYWORDS", () => {
|
||||
it("contains all required keywords", () => {
|
||||
const required = [
|
||||
"acceptance criteria",
|
||||
"done when",
|
||||
"success criteria",
|
||||
"definition of done",
|
||||
"requirements:",
|
||||
"must:",
|
||||
"should:",
|
||||
"- [ ]",
|
||||
"- [x]",
|
||||
"deliverable:",
|
||||
"expected outcome",
|
||||
"exit criteria",
|
||||
];
|
||||
for (const kw of required) {
|
||||
expect(ACCEPTANCE_CRITERIA_KEYWORDS).toContain(kw);
|
||||
}
|
||||
});
|
||||
});
|
||||
|
||||
// --- TASK-03: checkCapabilityMatch tests ---
|
||||
|
||||
describe("checkCapabilityMatch", () => {
|
||||
function makeMockDb(agent: { name: string; capabilities: string }, skillAssignments: string[] = [], skills: Record<string, { name: string; description: string | null; slug: string }> = {}) {
|
||||
return {
|
||||
getAgent: (_id: string) => agent,
|
||||
listAgentSkillAssignments: (_id: string) => skillAssignments,
|
||||
getCompanySkill: (_companyId: string, skillId: string) => skills[skillId] ?? null,
|
||||
};
|
||||
}
|
||||
|
||||
it("returns matched=true when agent capabilities overlap with task", () => {
|
||||
const db = makeMockDb({ name: "Frontend Dev", capabilities: "javascript react frontend" });
|
||||
const result = checkCapabilityMatch(db, "agent-1", "Build React login form with validation", "company-1");
|
||||
expect(result.matched).toBe(true);
|
||||
});
|
||||
|
||||
it("returns matched=false with warning when capabilities do not overlap", () => {
|
||||
const db = makeMockDb({ name: "Data Analyst", capabilities: "python data analysis" });
|
||||
const result = checkCapabilityMatch(db, "agent-1", "Build React login form with validation", "company-1");
|
||||
expect(result.matched).toBe(false);
|
||||
expect(result.warning).toBeDefined();
|
||||
expect(result.warning).toContain("Data Analyst");
|
||||
});
|
||||
|
||||
it("returns matched=true when agent has empty capabilities and no skills (graceful degradation)", () => {
|
||||
const db = makeMockDb({ name: "New Agent", capabilities: "" });
|
||||
const result = checkCapabilityMatch(db, "agent-1", "Build React login form", "company-1");
|
||||
expect(result.matched).toBe(true);
|
||||
});
|
||||
|
||||
it("returns matched=true when agent has no capabilities but skill name overlaps with task", () => {
|
||||
const db = makeMockDb(
|
||||
{ name: "Specialist", capabilities: "" },
|
||||
["skill-1"],
|
||||
{ "skill-1": { name: "react development", description: "Building React components", slug: "react-dev" } },
|
||||
);
|
||||
const result = checkCapabilityMatch(db, "agent-1", "Build React component with hooks", "company-1");
|
||||
expect(result.matched).toBe(true);
|
||||
});
|
||||
|
||||
it("returns matched=true when task description has no significant words", () => {
|
||||
const db = makeMockDb({ name: "Agent", capabilities: "python" });
|
||||
const result = checkCapabilityMatch(db, "agent-1", "Do it", "company-1");
|
||||
// "do" and "it" are <= 3 chars, so taskWords is empty -> matched=true
|
||||
expect(result.matched).toBe(true);
|
||||
});
|
||||
});
|
||||
Reference in new issue
Block a user