Compare commits

..
Author SHA1 Message Date
SinabinaandClaude Fable 5 453c480989 fix(gateway): add chat touchpoint to zhipu recipe so GLM subagents work (#1157)
The zhipu recipe was embedding-only, so models.tier.subagent=zhipu:glm-5.1
threw "does not offer a chat touchpoint" — while the error hint falsely
listed zhipu (and dashscope/minimax, also embedding-only) among providers
with chat.

- zhipu recipe: add a chat touchpoint (glm-5.1 family, supports_tools +
  supports_subagent_loop; no Anthropic-style prompt cache on the
  OpenAI-compat path, so the loop runs with the degraded:no_caching warn).
  openai-compat tier means newer GLM ids pass without a recipe edit.
- capabilities.ts: compute the "Known providers with chat" hint from the
  recipe registry instead of a hardcoded list, so it can never drift into
  naming chat-less providers again.
- Declines the originally requested models.anthropic_compatible_prefixes
  config: v0.38's recipe-driven capability gate already replaced the
  Anthropic-only enforcement, so a recipe chat touchpoint is the whole fix.

Fixes #1157

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-21 14:22:03 -07:00
8 changed files with 75 additions and 138 deletions
+5 -8
View File
@@ -21,17 +21,14 @@ GBrain is tuned for the Supabase **Transaction pooler** (port 6543): it
auto-disables prepared statements there and routes `engine.transaction()`
(migrations, DDL, sync imports) to a derived **direct** connection
(`db.<ref>.supabase.co:5432`). That direct host is IPv6-only, so on an
IPv4-only host it is unreachable. When that happens gbrain now falls back to
the pooler automatically (one stderr warning, then single-pool mode for the
rest of the process) — but the pooler's ~2-min statement timeout can truncate
very long migrations or bulk imports.
IPv4-only host, reads work but sync **silently skips most pages**. This is the
number one cause of "sync ran but nothing happened."
Fix: make the direct connection reachable over IPv4. Either set
`GBRAIN_DIRECT_DATABASE_URL` to the **Session pooler** string (port 5432 on the
`pooler.supabase.com` host, IPv4), or enable Supabase's IPv4 add-on.
`GBRAIN_DISABLE_DIRECT_POOL=1` skips the direct pool (and the fallback warning)
entirely. Verify by running `gbrain sync` and checking that the page count in
`gbrain stats` matches the syncable file count in the repo.
`pooler.supabase.com` host, IPv4), or enable Supabase's IPv4 add-on. Verify by
running `gbrain sync` and checking that the page count in `gbrain stats` matches
the syncable file count in the repo.
### The Primitives
+5 -8
View File
@@ -2720,17 +2720,14 @@ GBrain is tuned for the Supabase **Transaction pooler** (port 6543): it
auto-disables prepared statements there and routes `engine.transaction()`
(migrations, DDL, sync imports) to a derived **direct** connection
(`db.<ref>.supabase.co:5432`). That direct host is IPv6-only, so on an
IPv4-only host it is unreachable. When that happens gbrain now falls back to
the pooler automatically (one stderr warning, then single-pool mode for the
rest of the process) — but the pooler's ~2-min statement timeout can truncate
very long migrations or bulk imports.
IPv4-only host, reads work but sync **silently skips most pages**. This is the
number one cause of "sync ran but nothing happened."
Fix: make the direct connection reachable over IPv4. Either set
`GBRAIN_DIRECT_DATABASE_URL` to the **Session pooler** string (port 5432 on the
`pooler.supabase.com` host, IPv4), or enable Supabase's IPv4 add-on.
`GBRAIN_DISABLE_DIRECT_POOL=1` skips the direct pool (and the fallback warning)
entirely. Verify by running `gbrain sync` and checking that the page count in
`gbrain stats` matches the syncable file count in the repo.
`pooler.supabase.com` host, IPv4), or enable Supabase's IPv4 add-on. Verify by
running `gbrain sync` and checking that the page count in `gbrain stats` matches
the syncable file count in the repo.
### The Primitives
-6
View File
@@ -1078,9 +1078,6 @@ async function initPostgres(opts: {
console.warn(' Direct connections are IPv6 only and fail in many environments.');
console.warn(' Use the Transaction pooler connection string instead (port 6543):');
console.warn(' Supabase Dashboard > Connect (top bar) > Connection String > Transaction pooler');
console.warn(' (With a pooler URL, gbrain derives a direct connection for DDL and falls back');
console.warn(' to the pooler automatically if that host is unreachable. Power users:');
console.warn(' GBRAIN_DIRECT_DATABASE_URL overrides the derived URL; GBRAIN_DISABLE_DIRECT_POOL=1 disables it.)');
console.warn('');
}
@@ -1094,9 +1091,6 @@ async function initPostgres(opts: {
if (databaseUrl.includes('supabase.co') && (msg.includes('ECONNREFUSED') || msg.includes('ETIMEDOUT'))) {
console.error('Connection failed. Supabase direct connections (db.*.supabase.co:5432) are IPv6 only.');
console.error('Use the Transaction pooler connection string instead (port 6543).');
console.error('(gbrain derives its own direct connection from pooler URLs for DDL; if that host is');
console.error('unreachable it falls back to the pooler. GBRAIN_DIRECT_DATABASE_URL overrides the');
console.error('derived URL; GBRAIN_DISABLE_DIRECT_POOL=1 disables the direct pool entirely.)');
}
throw e;
}
+5 -1
View File
@@ -22,6 +22,7 @@
*/
import { resolveRecipe } from './model-resolver.ts';
import { listRecipes } from './recipes/index.ts';
import { AIConfigError } from './errors.ts';
export interface ProviderCapabilities {
@@ -77,7 +78,10 @@ export function getProviderCapabilities(modelString: string): ProviderCapabiliti
if (!chat) {
throw new AIConfigError(
`Provider "${recipe.id}" does not offer a chat touchpoint.`,
`Known providers with chat: openai, anthropic, google, openrouter, litellm-proxy, deepseek, groq, together, azure-openai, dashscope, minimax, zhipu, ollama, llama-server. Pick one for models.tier.subagent.`,
// Computed from the registry so the hint can't drift into listing
// chat-less providers (the pre-fix list falsely included embedding-only
// recipes, sending users in circles — #1157).
`Known providers with chat: ${listRecipes().filter(r => r.touchpoints.chat).map(r => r.id).join(', ')}. Pick one for models.tier.subagent.`,
);
}
+19 -4
View File
@@ -1,9 +1,10 @@
import type { Recipe } from '../types.ts';
/**
* Zhipu AI (智谱AI) BigModel Open Platform. OpenAI-compatible /embeddings
* endpoint at open.bigmodel.cn. Hosts embedding-2 (1024d) and embedding-3
* (Matryoshka up to 2048d).
* Zhipu AI (智谱AI) BigModel Open Platform. OpenAI-compatible /embeddings and
* /chat/completions endpoints at open.bigmodel.cn. Hosts embedding-2 (1024d),
* embedding-3 (Matryoshka up to 2048d), and the GLM chat family (glm-5.1 etc.)
* with native tool calling — usable for models.tier.subagent (#1157).
*
* embedding-3 at 2048 dims exceeds pgvector's HNSW cap of 2000 — those
* brains fall back to exact vector scans (see
@@ -25,6 +26,20 @@ export const zhipu: Recipe = {
setup_url: 'https://open.bigmodel.cn/',
},
touchpoints: {
chat: {
// Informational list (openai-compat tier: assertTouchpoint doesn't
// enforce it), so newer GLM ids pass without a recipe edit.
models: ['glm-5.1', 'glm-4.6', 'glm-4.5'],
supports_tools: true,
// gbrain-side stable tool ids (v0.38 D11) decoupled the loop from
// Anthropic response formats; GLM tool calling is stable through the
// OpenAI-compat path, same as deepseek/groq.
supports_subagent_loop: true,
// Anthropic-style cache_control markers are not honored on the
// OpenAI-compat path — the loop runs hot (degraded:no_caching warn).
supports_prompt_cache: false,
max_context_tokens: 128000,
},
embedding: {
models: ['embedding-3', 'embedding-2'],
default_dims: 1024,
@@ -36,5 +51,5 @@ export const zhipu: Recipe = {
},
},
setup_hint:
'Get an API key at https://open.bigmodel.cn/, then `export ZHIPUAI_API_KEY=...`',
'Get an API key at https://open.bigmodel.cn/, then `export ZHIPUAI_API_KEY=...`. Chat/subagent: use `zhipu:glm-5.1`.',
};
+2 -48
View File
@@ -167,25 +167,6 @@ export function deriveDirectUrl(url: string): string | null {
}
}
/**
* Error codes that mean "the direct host is unreachable from this network"
* (#1641). The auto-derived db.<ref>.supabase.co host is IPv6-only without
* the paid IPv4 add-on, so ENOTFOUND/ECONNREFUSED here is expected on
* IPv4-only networks — we fall back to the pooler instead of failing init.
*/
const NETWORK_UNREACHABLE_CODES = [
'ENOTFOUND', 'ECONNREFUSED', 'ENETUNREACH', 'EHOSTUNREACH',
'ETIMEDOUT', 'CONNECT_TIMEOUT',
];
/** True when err looks like a network-unreachable failure (not auth/SQL). */
export function isNetworkUnreachableError(err: unknown): boolean {
const code = (err as { code?: unknown } | null)?.code;
if (typeof code === 'string' && NETWORK_UNREACHABLE_CODES.includes(code)) return true;
const msg = err instanceof Error ? err.message : String(err);
return NETWORK_UNREACHABLE_CODES.some(c => msg.includes(c));
}
/**
* Read kill-switch state from env. Subordinate to parent manager's state
* when present (A2 inheritance).
@@ -338,30 +319,7 @@ export class ConnectionManager {
throw err;
});
}
let pool: Sql | null;
try {
pool = await this._directInit;
} catch (err) {
// #1641: the derived direct host (db.<ref>.supabase.co) is IPv6-only
// without Supabase's IPv4 add-on. On IPv4-only networks the direct
// pool can never connect — permanently fall back to the read pool
// (self-activating kill-switch) instead of failing init/migrations.
// Non-network errors (auth, SQL) still throw: they mean misconfig,
// not unreachability.
if (isNetworkUnreachableError(err)) {
const alreadyWarned = this._killSwitch;
this._killSwitch = true;
const msg = err instanceof Error ? err.message : String(err);
if (!alreadyWarned) console.error(
`gbrain: direct connection to ${this._directUrl ? this.hostOnly(this._directUrl) : 'unknown host'} unreachable (${msg}); ` +
'falling back to the pooler for DDL/bulk (long migrations may hit the pooler statement timeout). ' +
'Set GBRAIN_DIRECT_DATABASE_URL to a reachable direct URL (e.g. the Session pooler, port 5432) or enable the Supabase IPv4 add-on; ' +
'GBRAIN_DISABLE_DIRECT_POOL=1 silences this.',
);
return this.getReadPool();
}
throw err;
}
const pool = await this._directInit;
if (!pool) {
// Defensive — initDirectPool should have thrown.
throw new Error('connection-manager: direct pool init returned null');
@@ -392,9 +350,8 @@ export class ConnectionManager {
},
};
const t0 = Date.now();
let pool: Sql | null = null;
try {
pool = postgres(this._directUrl, opts);
const pool = postgres(this._directUrl, opts);
// Probe to validate connectivity early.
await pool`SELECT 1`;
logConnectionEvent({
@@ -405,9 +362,6 @@ export class ConnectionManager {
});
return pool;
} catch (err) {
// Don't leak the failed pool's sockets/timers (#1641 fallback keeps
// the process running afterward).
if (pool) await endPoolBounded(pool);
logConnectionEvent({
pool: 'ddl',
op: 'error',
+39
View File
@@ -69,6 +69,45 @@ describe('recipe: zhipu', () => {
expect(sql.toLowerCase()).toContain('hnsw');
});
test('chat touchpoint declares GLM models with tool + subagent-loop support (#1157)', () => {
const r = getRecipe('zhipu')!;
expect(r.touchpoints.chat).toBeDefined();
expect(r.touchpoints.chat!.models).toContain('glm-5.1');
expect(r.touchpoints.chat!.supports_tools).toBe(true);
expect(r.touchpoints.chat!.supports_subagent_loop).toBe(true);
expect(r.touchpoints.chat!.supports_prompt_cache).toBe(false);
});
test('zhipu:glm-5.1 passes the subagent capability gate (degraded:no_caching, not refused)', async () => {
// Pre-fix: getProviderCapabilities threw "does not offer a chat touchpoint"
// and classifyCapabilities returned 'unknown' → subagent submit refused.
const { getProviderCapabilities, classifyCapabilities } =
await import('../../src/core/ai/capabilities.ts');
const caps = getProviderCapabilities('zhipu:glm-5.1');
expect(caps.supportsToolCalling).toBe(true);
expect(classifyCapabilities('zhipu:glm-5.1')).toBe('degraded:no_caching');
});
test('no-chat-touchpoint error hint lists only providers that actually have chat', async () => {
// The hint is computed from the registry; every provider it names must
// really carry a chat touchpoint (pre-fix it hardcoded zhipu/dashscope/
// minimax, all embedding-only at the time).
const { getProviderCapabilities } = await import('../../src/core/ai/capabilities.ts');
const { listRecipes } = await import('../../src/core/ai/recipes/index.ts');
let hint = '';
try {
getProviderCapabilities('voyage:voyage-3');
throw new Error('expected AIConfigError for embedding-only provider');
} catch (e) {
hint = (e as { fix?: string }).fix ?? String(e);
}
const listed = hint.match(/chat: ([^.]+)\./)?.[1]?.split(', ') ?? [];
expect(listed.length).toBeGreaterThan(0);
const withChat = new Set(listRecipes().filter(r => r.touchpoints.chat).map(r => r.id));
for (const id of listed) expect(withChat.has(id)).toBe(true);
expect(listed).toContain('zhipu');
});
test('dimsProviderOptions threads dimensions for embedding-3 (Matryoshka)', async () => {
// Codex finding #1: Zhipu embedding-3 is Matryoshka 256-2048. Without
// `dimensions` on the wire, user-selected non-default dims are
-63
View File
@@ -3,7 +3,6 @@ import {
isSupabasePoolerUrl,
deriveDirectUrl,
readKillSwitchEnv,
isNetworkUnreachableError,
resolveDirectPoolSize,
ConnectionManager,
DEFAULT_DIRECT_POOL_SIZE,
@@ -239,65 +238,3 @@ describe('ConnectionManager — parent inheritance (A2)', () => {
}
});
});
describe('isNetworkUnreachableError (#1641)', () => {
test('classifies network codes as unreachable', () => {
for (const code of ['ENOTFOUND', 'ECONNREFUSED', 'ENETUNREACH', 'EHOSTUNREACH', 'ETIMEDOUT', 'CONNECT_TIMEOUT']) {
const err = Object.assign(new Error('connect failed'), { code });
expect(isNetworkUnreachableError(err)).toBe(true);
}
});
test('classifies by message when code absent', () => {
expect(isNetworkUnreachableError(new Error('getaddrinfo ENOTFOUND db.abc.supabase.co'))).toBe(true);
});
test('auth/SQL errors are NOT unreachable', () => {
expect(isNetworkUnreachableError(new Error('password authentication failed for user "postgres"'))).toBe(false);
expect(isNetworkUnreachableError(new Error('syntax error at or near "SELEC"'))).toBe(false);
expect(isNetworkUnreachableError(null)).toBe(false);
});
});
describe('ConnectionManager — direct-pool fallback on unreachable host (#1641)', () => {
let originalKillSwitch: string | undefined;
let originalError: typeof console.error;
let errLines: string[];
beforeEach(() => {
originalKillSwitch = process.env.GBRAIN_DISABLE_DIRECT_POOL;
delete process.env.GBRAIN_DISABLE_DIRECT_POOL;
originalError = console.error;
errLines = [];
console.error = (...args: unknown[]) => { errLines.push(args.join(' ')); };
});
afterEach(() => {
console.error = originalError;
if (originalKillSwitch === undefined) delete process.env.GBRAIN_DISABLE_DIRECT_POOL;
else process.env.GBRAIN_DISABLE_DIRECT_POOL = originalKillSwitch;
});
test('ddl() falls back to the read pool when the direct host is unreachable', async () => {
const cm = new ConnectionManager({
url: 'postgresql://postgres.abc:p@aws.pooler.supabase.com:6543/db',
// 127.0.0.1:9 (discard) → instant ECONNREFUSED, the IPv4-only-network shape.
directUrl: 'postgresql://postgres:p@127.0.0.1:9/db',
});
const fakeReadPool = {} as ReturnType<typeof ConnectionManager.prototype.read>;
cm.setReadPool(fakeReadPool);
expect(cm.isDualPoolActive()).toBe(true);
const pool = await cm.ddl(); // without the fix this throws ECONNREFUSED
expect(pool).toBe(fakeReadPool);
// Self-activating kill-switch: subsequent calls skip the direct pool.
expect(cm.isKillSwitchActive()).toBe(true);
expect(cm.isDualPoolActive()).toBe(false);
expect(cm.describeMode().mode).toBe('single (kill-switch)');
// One stderr line mentioning the power-user override.
const warning = errLines.filter(l => l.includes('GBRAIN_DIRECT_DATABASE_URL'));
expect(warning.length).toBe(1);
const again = await cm.ddl();
expect(again).toBe(fakeReadPool);
expect(errLines.filter(l => l.includes('GBRAIN_DIRECT_DATABASE_URL')).length).toBe(1);
}, 20000);
});