Skip to content

zcode 0.1.40: freeFirst's refusals show on the card, and a blocked address rests the whole gift #545

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
jinpy666 wants to merge 3 commits into
base: main
Choose a base branch
from
+251 −14
Open

zcode 0.1.40: freeFirst's refusals show on the card, and a blocked address rests the whole gift #545

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
0d2c15c
zcode 0.1.40: freeFirst's refusals show on the card, and a blocked ad…
jinpy666 Oct 31, 2026
e717472
zcode: the card and the -Trial refusal say which kind of block it is
jinpy666 Oct 31, 2026
346e902
zcode: a blocked or restricted -Trial refusal is not a rate limit
jinpy666 Oct 31, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 18 additions & 2 deletions packages/zcode/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -168,9 +168,25 @@ behind it go straight to the plan. A gift that cannot be asked at all —
the network, the host — is answered from the plan the same way, the
route read dropped for the next request, while a caller's own abort
stays an abort. The agent sees the plan's answer
either way. And the `-Trial` entries keep their own strict semantics —
either way. When a refusal is fresh, the usage card says so — **"ZCode's
gift is refusing requests — freeFirst answers from the coding plan until
14:02 (…)"** — so a gift sitting idle
beside a fat balance is explained, not silent. The why tells the two
block kinds apart: a block page from the edge firewall says "the
network's address is blocked" (every model the gift serves rests for the
route's ten minutes — an address block outlasts a minute), ZCode's own
3012 says "ZCode restricted this account", a spent answer rests its one
model for a minute as "its quota said spent", and any other failure is
named with its status; while a rest is on, the
requests behind it go straight to the plan. The `-Trial` entries keep
their own strict semantics —
asking one is still an explicit pick of the gift alone, never the plan;
a spent one answers 429 with freeFirst on, as it does without it.
a spent one answers 429 with freeFirst on, as it does without it, and
one asked while the gift is blocked or restricted says that with the
block's own hint ("…the provider's network firewall blocked requests
from this IP…" / "…ZCode's risk control blocked this account…") and an
error type that is not a rate limit, so the gateway rests the route by
what happened; a model the gift never serves keeps its no-quota 400.

## A few at once

Expand Down
169 changes: 169 additions & 0 deletions packages/zcode/dual.test.mjs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,6 +36,7 @@ const real = globalThis.fetch
beforeEach(async () => {
_internal.routes.clear()
_internal.blocked.clear()
_internal.refused.clear()
sent.length = 0
for (const k of Object.keys(answers)) delete answers[k]
globalThis.fetch = async (input, init = {}) => {
Expand Down Expand Up @@ -386,6 +387,174 @@ test("an abort while the gift is in flight stays an abort", async () => {
expect(tries.plan).toBe(0)
})

test("a firewall block on the gift shows on the card and rests the gift's models", async () => {
planAndBalance()
answers["/api/monitor/usage/quota/limit"] = ok({ limits: [{ type: "CREDIT_LIMIT", unit: 6, number: 1, percentage: 20 }] })
answers["/api/v1/zcode-plan/anthropic/v1/messages"] = new Response('<!doctypehtml><html lang="zh-cn"><title>405</title>errors.aliyun.com</html>', { status: 405 })
answers["/api/anthropic/v1/messages"] = ok({ type: "message", model: "GLM-5.3-Flash", content: [] })
const h = await freePlugin({ freeFirst: true })
const opts = await h.loader(async () => coding, { id: "zcode" })
const res = await opts.fetch(MSG, { method: "POST", headers: { "content-type": "application/json" }, body: body("GLM-5.3-Flash") })
expect(res.status).toBe(200)
expect(JSON.parse(sent.at(-1).body).model).toBe("GLM-5.3-Flash")
// while the block is fresh the gift is skipped entirely: no gift try,
// the plan answers straight
const before = sent.filter((c) => c.origin === "https://zcode.z.ai" && c.path.endsWith("/messages")).length
await opts.fetch(MSG, { method: "POST", headers: { "content-type": "application/json" }, body: body("GLM-5.3-Flash") })
expect(sent.filter((c) => c.origin === "https://zcode.z.ai" && c.path.endsWith("/messages")).length).toBe(before)
// the card says why the gift sits idle beside its balance
const u = await h.usage(async () => coding, { id: "zcode" })
const line = (u.windows ?? []).find((w) => w.aside && /refusing requests/.test(w.name))
expect(line.display).toMatch(/coding plan until \d\d:\d\d/)
expect(line.display).toContain("blocked")
})

test("a spent answer names itself on the card for its minute", async () => {
planAndBalance()
answers["/api/monitor/usage/quota/limit"] = ok({ limits: [{ type: "CREDIT_LIMIT", unit: 6, number: 1, percentage: 20 }] })
answers["/api/v1/zcode-plan/anthropic/v1/messages"] = new Response(JSON.stringify({ error: { message: "Insufficient balance or no resource package", code: 1113 } }), { status: 429, headers: { "content-type": "application/json" } })
answers["/api/anthropic/v1/messages"] = ok({ type: "message", content: [] })
const h = await freePlugin({ freeFirst: true })
const opts = await h.loader(async () => coding, { id: "zcode" })
const res = await opts.fetch(MSG, { method: "POST", headers: { "content-type": "application/json" }, body: body("GLM-5.3-Flash") })
expect(res.status).toBe(200)
const u = await h.usage(async () => coding, { id: "zcode" })
const line = (u.windows ?? []).find((w) => w.aside && /refusing requests/.test(w.name))
expect(line.display).toMatch(/coding plan until \d\d:\d\d/)
expect(line.display).toContain("spent")
})

test("a gift failure the plan answers shows on the card with the status", async () => {
planAndBalance()
answers["/api/monitor/usage/quota/limit"] = ok({ limits: [{ type: "CREDIT_LIMIT", unit: 6, number: 1, percentage: 20 }] })
answers["/api/v1/zcode-plan/anthropic/v1/messages"] = new Response("boom", { status: 500 })
answers["/api/anthropic/v1/messages"] = ok({ type: "message", content: [] })
const h = await freePlugin({ freeFirst: true })
const opts = await h.loader(async () => coding, { id: "zcode" })
const res = await opts.fetch(MSG, { method: "POST", headers: { "content-type": "application/json" }, body: body("GLM-5.3-Flash") })
expect(res.status).toBe(200)
const u = await h.usage(async () => coding, { id: "zcode" })
const line = (u.windows ?? []).find((w) => w.aside && /refusing requests/.test(w.name))
expect(line.display).toMatch(/coding plan until \d\d:\d\d/)
expect(line.display).toContain("500")
})

test("a 3012 says the account is restricted, not the address", async () => {
planAndBalance()
answers["/api/monitor/usage/quota/limit"] = ok({ limits: [{ type: "CREDIT_LIMIT", unit: 6, number: 1, percentage: 20 }] })
answers["/api/v1/zcode-plan/anthropic/v1/messages"] = new Response(JSON.stringify({ error: { message: "request has been blocked due to unusual activity", code: 3012 } }), { status: 405, headers: { "content-type": "application/json" } })
answers["/api/anthropic/v1/messages"] = ok({ type: "message", model: "GLM-5.3-Flash", content: [] })
const h = await freePlugin({ freeFirst: true })
const opts = await h.loader(async () => coding, { id: "zcode" })
const res = await opts.fetch(MSG, { method: "POST", headers: { "content-type": "application/json" }, body: body("GLM-5.3-Flash") })
expect(res.status).toBe(200)
const u = await h.usage(async () => coding, { id: "zcode" })
const line = (u.windows ?? []).find((w) => w.aside && /refusing requests/.test(w.name))
expect(line.display).toContain("restricted")
expect(line.display).not.toContain("blocked")
})

test("a blocked gift's -Trial entry says the network is blocked, not the quota", async () => {
planAndBalance()
answers["/api/monitor/usage/quota/limit"] = ok({ limits: [{ type: "CREDIT_LIMIT", unit: 6, number: 1, percentage: 20 }] })
answers["/api/v1/zcode-plan/anthropic/v1/messages"] = new Response('<!doctypehtml><html lang="zh-cn"><title>405</title>errors.aliyun.com</html>', { status: 405 })
answers["/api/anthropic/v1/messages"] = ok({ type: "message", content: [] })
const h = await freePlugin({ freeFirst: true })
const opts = await h.loader(async () => coding, { id: "zcode" })
await opts.fetch(MSG, { method: "POST", headers: { "content-type": "application/json" }, body: body("GLM-5.3-Flash") })
const res = await opts.fetch(MSG, { method: "POST", headers: { "content-type": "application/json" }, body: body("GLM-5.3-Flash-Trial") })
expect(res.status).toBe(429)
const said = await res.json()
// an address block is not a rate limit: magpie classifies by the type
expect(said.error.type).toBe("invalid_request_error")
expect(said.error.message).toContain("network firewall blocked requests from this IP")
expect(said.error.message).not.toContain("quota")
})

test("a restricted gift's -Trial entry says the account is restricted", async () => {
planAndBalance()
answers["/api/monitor/usage/quota/limit"] = ok({ limits: [{ type: "CREDIT_LIMIT", unit: 6, number: 1, percentage: 20 }] })
answers["/api/v1/zcode-plan/anthropic/v1/messages"] = new Response(JSON.stringify({ error: { message: "request has been blocked due to unusual activity", code: 3012 } }), { status: 405, headers: { "content-type": "application/json" } })
answers["/api/anthropic/v1/messages"] = ok({ type: "message", content: [] })
const h = await freePlugin({ freeFirst: true })
const opts = await h.loader(async () => coding, { id: "zcode" })
await opts.fetch(MSG, { method: "POST", headers: { "content-type": "application/json" }, body: body("GLM-5.3-Flash") })
const res = await opts.fetch(MSG, { method: "POST", headers: { "content-type": "application/json" }, body: body("GLM-5.3-Flash-Trial") })
expect(res.status).toBe(429)
const said = await res.json()
expect(said.error.type).toBe("invalid_request_error")
expect(said.error.message).toContain("risk control blocked this account")
})

test("a spent gift's -Trial entry keeps its quota reply", async () => {
planAndBalance()
answers["/api/monitor/usage/quota/limit"] = ok({ limits: [{ type: "CREDIT_LIMIT", unit: 6, number: 1, percentage: 20 }] })
// the freeFirst plain request hears the gift say its quota is spent, and
// the trial entry after it is answered from that rest, locally
answers["/api/v1/zcode-plan/anthropic/v1/messages"] = new Response(JSON.stringify({ error: { message: "Insufficient balance or no resource package", code: 1113 } }), { status: 429, headers: { "content-type": "application/json" } })
answers["/api/anthropic/v1/messages"] = ok({ type: "message", content: [] })
const h = await freePlugin({ freeFirst: true })
const opts = await h.loader(async () => coding, { id: "zcode" })
await opts.fetch(MSG, { method: "POST", headers: { "content-type": "application/json" }, body: body("GLM-5.3-Flash") })
const res = await opts.fetch(MSG, { method: "POST", headers: { "content-type": "application/json" }, body: body("GLM-5.3-Flash-Trial") })
expect(res.status).toBe(429)
const said = await res.json()
expect(said.error.type).toBe("rate_limit_error")
expect(said.error.message).toContain("limited quota")
})

test("a -Trial entry the gift never serves keeps its no-quota 400", async () => {
planAndBalance()
answers["/api/monitor/usage/quota/limit"] = ok({ limits: [{ type: "CREDIT_LIMIT", unit: 6, number: 1, percentage: 20 }] })
answers["/api/v1/zcode-plan/anthropic/v1/messages"] = new Response('<!doctypehtml><html lang="zh-cn"><title>405</title>errors.aliyun.com</html>', { status: 405 })
answers["/api/anthropic/v1/messages"] = ok({ type: "message", content: [] })
const h = await freePlugin({ freeFirst: true })
const opts = await h.loader(async () => coding, { id: "zcode" })
await opts.fetch(MSG, { method: "POST", headers: { "content-type": "application/json" }, body: body("GLM-5.3-Flash") })
const res = await opts.fetch(MSG, { method: "POST", headers: { "content-type": "application/json" }, body: body("GLM-5.2-Trial") })
expect(res.status).toBe(400)
const said = await res.json()
expect(said.error.type).toBe("invalid_request_error")
expect(said.error.message).toContain("no active limited quota")
expect(said.error.message).not.toContain("blocked")
})

test("a gift success clears the card's refusal line", async () => {
planAndBalance()
answers["/api/monitor/usage/quota/limit"] = ok({ limits: [{ type: "CREDIT_LIMIT", unit: 6, number: 1, percentage: 20 }] })
answers["/api/v1/zcode-plan/anthropic/v1/messages"] = new Response('<!doctypehtml><html lang="zh-cn"><title>405</title>errors.aliyun.com</html>', { status: 405 })
answers["/api/anthropic/v1/messages"] = ok({ type: "message", content: [] })
const h = await freePlugin({ freeFirst: true })
const opts = await h.loader(async () => coding, { id: "zcode" })
await opts.fetch(MSG, { method: "POST", headers: { "content-type": "application/json" }, body: body("GLM-5.3-Flash") })
// the block out, the gift answers: the line goes with it
answers["/api/v1/zcode-plan/anthropic/v1/messages"] = ok({ type: "message", content: [] })
_internal.blocked.set(KEY + "\0" + jwt + "\0glm-5.3-flash", Date.now() - 1)
await opts.fetch(MSG, { method: "POST", headers: { "content-type": "application/json" }, body: body("GLM-5.3-Flash") })
const u = await h.usage(async () => coding, { id: "zcode" })
expect((u.windows ?? []).find((w) => w.aside && /refusing requests/.test(w.name))).toBeUndefined()
})

test("the gift's own busy cap is not a refusal for the card", async () => {
planAndBalance()
answers["/api/monitor/usage/quota/limit"] = ok({ limits: [{ type: "CREDIT_LIMIT", unit: 6, number: 1, percentage: 20 }] })
const waits = _internal.BUSY_WAITS.splice(0)
_internal.BUSY_WAITS.push(1)
try {
answers["/api/v1/zcode-plan/anthropic/v1/messages"] = new Response(JSON.stringify({ code: "3008", msg: "user concurrency limit exceeded" }), { status: 429, headers: { "content-type": "application/json" } })
answers["/api/anthropic/v1/messages"] = ok({ type: "message", content: [] })
const h = await freePlugin({ freeFirst: true })
const opts = await h.loader(async () => coding, { id: "zcode" })
const res = await opts.fetch(MSG, { method: "POST", headers: { "content-type": "application/json" }, body: body("GLM-5.3-Flash") })
expect(res.status).toBe(200)
const u = await h.usage(async () => coding, { id: "zcode" })
expect((u.windows ?? []).find((w) => w.aside && /refusing requests/.test(w.name))).toBeUndefined()
} finally {
_internal.BUSY_WAITS.length = 0
_internal.BUSY_WAITS.push(...waits)
}
})

for (const [name, paid, gift] of [
["a full paid week leaves an available gift usable", 100, 25],
["a spent gift leaves the paid plan usable", 20, 100],
Expand Down
Loading
Loading