mirror of
https://github.com/CopilotKit/CopilotKit.git
synced 2026-09-14 16:26:20 +08:00
84dd86f2ed
The marked block that wires managed Intelligence is the region a hosted reader copies verbatim, and nothing checked it. Both gaps were deliberate: the parity manifest lists `src/app/api/copilotkit/**` under `allowedDivergence` for every instance it tracks, and no `docker-compose.test.yml` sets `COPILOTKIT_LICENSE_TOKEN`, so every smoke-tested starter takes the else arm and the `intelligence:` arm has never run in CI. The cost was already visible. The block's code was byte-identical in 21 of 22 starters, but its warning comment had drifted into five variants and the two `ms-agent-framework-*` starters shipped the `demo-user` stub with no warning at all. That drift is how the localhost default of OSS-981 survived in all 22 copies at once. Add `scripts/validate-intelligence-wiring-block.ts`, which greps the opening marker, compares every site against the north-star starter, and fails on the first line that differs. Two normalisations keep it usable: the block is dedented, because `agentcore` nests it deeper, and the else arm's runner name is masked, because `agentcore` runs `AgentCoreRunner` in front of a Bedrock session where an in-process runner has nothing to run. Everything else, comment text included, must match to the byte. Then unify the warning at all 22 sites on the fullest wording, which also says the id must exist in Intelligence or thread operations can fail. The check passes on day one, so it is a ratchet rather than a migration. It is a shape gate, not a content gate: 22 identically wrong copies still pass. What it guarantees is that a fix reaches all of them or none. Not covered: enrolling the `intelligence:` arm in the smoke path. That needs a license token in CI and a reachable endpoint from the compose network, and is tracked separately.
180 lines
5.9 KiB
TypeScript
180 lines
5.9 KiB
TypeScript
import { describe, expect, it } from "vitest";
|
|
import {
|
|
OPEN_MARKER,
|
|
blockDiff,
|
|
extractBlock,
|
|
findViolations,
|
|
markerFiles,
|
|
maskRunner,
|
|
normalizeBlock,
|
|
runnerName,
|
|
} from "../validate-intelligence-wiring-block.js";
|
|
|
|
/**
|
|
* A starter's Intelligence wiring is the one region a hosted reader copies
|
|
* verbatim, and until this check landed nothing held it still. The parity
|
|
* manifest lists `src/app/api/copilotkit/**` under `allowedDivergence` for
|
|
* every instance it tracks, and no smoke test sets `COPILOTKIT_LICENSE_TOKEN`,
|
|
* so the `intelligence:` arm has never executed in CI (OSS-982).
|
|
*
|
|
* The block was already byte-identical in 21 of 22 starters when the check was
|
|
* written, so this is a ratchet rather than a migration. What had drifted was
|
|
* the warning comment — into five variants, two of them missing outright — and
|
|
* that drift is how the localhost default of OSS-981 survived in all 22 copies.
|
|
*/
|
|
|
|
const CANONICAL = ` // --- copilotkit:intelligence (remove this block to opt out) ---
|
|
...(process.env.COPILOTKIT_LICENSE_TOKEN
|
|
? {
|
|
intelligence: new CopilotKitIntelligence({
|
|
apiKey: process.env.INTELLIGENCE_API_KEY ?? "",
|
|
}),
|
|
identifyUser: () => ({ id: "demo-user", name: "Demo User" }),
|
|
licenseToken: process.env.COPILOTKIT_LICENSE_TOKEN,
|
|
}
|
|
: { runner: new InMemoryAgentRunner() }),
|
|
// --- /copilotkit:intelligence ---`;
|
|
|
|
describe("extractBlock", () => {
|
|
it("returns the marker-delimited region, both markers included", () => {
|
|
const source = `export const runtime = new CopilotRuntime({\n${CANONICAL}\n});\n`;
|
|
|
|
expect(extractBlock(source)).toBe(CANONICAL);
|
|
});
|
|
|
|
it("returns null when the file carries no opening marker", () => {
|
|
expect(
|
|
extractBlock("export const runtime = new CopilotRuntime({});\n"),
|
|
).toBe(null);
|
|
});
|
|
|
|
it("returns null for an unterminated block, so the caller can report it", () => {
|
|
const source = `${OPEN_MARKER}\n ...(process.env.COPILOTKIT_LICENSE_TOKEN ? {} : {}),\n`;
|
|
|
|
expect(extractBlock(source)).toBe(null);
|
|
});
|
|
});
|
|
|
|
/**
|
|
* The block sits at a different nesting depth in `agentcore`, whose runtime is
|
|
* a Lambda handler rather than a Next.js route, so a raw string comparison
|
|
* would report every line of it. Depth is not drift; the code is what matters.
|
|
*/
|
|
describe("normalizeBlock", () => {
|
|
it("dedents to the shallowest line", () => {
|
|
expect(normalizeBlock(" a\n b\n c")).toBe("a\n b\nc");
|
|
});
|
|
|
|
it("makes the same block at two nesting depths compare equal", () => {
|
|
const deeper = CANONICAL.split("\n")
|
|
.map((line) => ` ${line}`)
|
|
.join("\n");
|
|
|
|
expect(normalizeBlock(deeper)).toBe(normalizeBlock(CANONICAL));
|
|
});
|
|
|
|
it("strips carriage returns and trailing spaces", () => {
|
|
expect(normalizeBlock(" a \r\n b\r\n")).toBe("a\nb");
|
|
});
|
|
});
|
|
|
|
/**
|
|
* The else arm is the one line that legitimately differs: `agentcore` runs
|
|
* `AgentCoreRunner` because its agent is a Bedrock AgentCore session, not an
|
|
* in-process runner. Masking the name lets the rest of the block be compared
|
|
* exactly while the name itself is checked against a per-starter expectation.
|
|
*/
|
|
describe("runnerName", () => {
|
|
it("reads the runner out of the else arm", () => {
|
|
expect(runnerName(CANONICAL)).toBe("InMemoryAgentRunner");
|
|
});
|
|
|
|
it("reads a different runner", () => {
|
|
const agentcore = CANONICAL.replace(
|
|
"InMemoryAgentRunner",
|
|
"AgentCoreRunner",
|
|
);
|
|
|
|
expect(runnerName(agentcore)).toBe("AgentCoreRunner");
|
|
});
|
|
|
|
it("returns null when the else arm constructs nothing", () => {
|
|
const noRunner = CANONICAL.replace(
|
|
": { runner: new InMemoryAgentRunner() }),",
|
|
": {}),",
|
|
);
|
|
|
|
expect(runnerName(noRunner)).toBe(null);
|
|
});
|
|
});
|
|
|
|
describe("maskRunner", () => {
|
|
it("makes two blocks that differ only in the runner compare equal", () => {
|
|
const agentcore = CANONICAL.replace(
|
|
"InMemoryAgentRunner",
|
|
"AgentCoreRunner",
|
|
);
|
|
|
|
expect(maskRunner(agentcore)).toBe(maskRunner(CANONICAL));
|
|
});
|
|
|
|
it("leaves every other difference visible", () => {
|
|
const drifted = CANONICAL.replace("demo-user", "someone-else");
|
|
|
|
expect(maskRunner(drifted)).not.toBe(maskRunner(CANONICAL));
|
|
});
|
|
});
|
|
|
|
describe("blockDiff", () => {
|
|
it("returns null for identical blocks", () => {
|
|
expect(blockDiff(CANONICAL, CANONICAL)).toBe(null);
|
|
});
|
|
|
|
it("reports the first differing line, numbered from one", () => {
|
|
const drifted = CANONICAL.replace(
|
|
' identifyUser: () => ({ id: "demo-user", name: "Demo User" }),',
|
|
' identifyUser: () => ({ id: "someone-else", name: "Demo User" }),',
|
|
);
|
|
|
|
expect(blockDiff(CANONICAL, drifted)).toEqual({
|
|
line: 7,
|
|
expected:
|
|
' identifyUser: () => ({ id: "demo-user", name: "Demo User" }),',
|
|
actual:
|
|
' identifyUser: () => ({ id: "someone-else", name: "Demo User" }),',
|
|
});
|
|
});
|
|
|
|
it("reports a missing line as an absent actual", () => {
|
|
const truncated = CANONICAL.split("\n").slice(0, 3).join("\n");
|
|
|
|
expect(blockDiff(CANONICAL, truncated)?.line).toBe(4);
|
|
});
|
|
|
|
it("reports an extra line as an absent expectation", () => {
|
|
const extended = `${CANONICAL}\n // trailing`;
|
|
|
|
expect(blockDiff(CANONICAL, extended)).toEqual({
|
|
line: 12,
|
|
expected: null,
|
|
actual: " // trailing",
|
|
});
|
|
});
|
|
});
|
|
|
|
/**
|
|
* Two assertions, because either alone can pass while the check does nothing.
|
|
* A count that dropped to zero would make the violation list vacuously empty —
|
|
* the failure mode of every `passWithNoTests` gate — so the file count is
|
|
* asserted against the starter inventory as well.
|
|
*/
|
|
describe("the repository's wiring sites", () => {
|
|
it("finds one marked site in every starter that wires Intelligence", () => {
|
|
expect(markerFiles().length).toBeGreaterThanOrEqual(22);
|
|
});
|
|
|
|
it("holds every site to one shape", () => {
|
|
expect(findViolations()).toEqual([]);
|
|
});
|
|
});
|