ZacharyZhang-NY/Kigi-CLI
1
0
Fork 0
Code Issues Pull Requests Actions 13 Packages Projects Releases Wiki Activity

feat(providers): add xAI (Grok) + migrate house BYOK env to KIGI_API_KEY

Browse Source 
13th provider (16th registry variant). Also reconciles a naming collision
the matrix flagged: this fork is house-branded "xai" (cf. xai.dev metadata,
KIGI_CODE_XAI_API_KEY legacy env), so XAI_API_KEY + method xai.api_key were
the GENERIC house BYOK, not x.ai/Grok. The provider table wants xai/XAI_API_KEY
for Grok.

Resolution (user-approved): XAI_API_KEY now keys the x.ai/Grok provider; the
house BYOK primary env moves to KIGI_API_KEY, keeping XAI_API_KEY and
KIGI_CODE_XAI_API_KEY as back-compat fallbacks (read_xai_api_key_env checks
KIGI_API_KEY first). The xai.api_key method id is unchanged (persisted-session
compat); the platform method id is the bare "xai", distinct from it.

xAI spec: api.x.ai/v1, Bearer, OpenAI listing + ChatCompletions, Passthrough
(docs confirm stream_options.include_usage accepted). /v1/models is minimal
(ids only) and requires auth, so it doubles as the key validator (401 on bad
key, no override) and metadata comes from models.dev enrichment. Live ids
match the models.dev "xai" keys byte-for-byte, so restrict_to_enriched keeps
the 5 tool-calling chat models (grok-4.5/4.3/4.20-0309-*/build-0.1) and drops
the grok-imagine-* generators + the non-tool multi-agent model. Snapshot
regenerated to include the xai provider (was stale; gen script already listed
it in TARGETS).

Env migration is comprehensive to avoid keying the xai platform (which would
trigger a live api.x.ai fetch) or leaving house-key reads stranded: routed
the trace CLI resolver + acp_agent/auth.json bridge + paste-key ext handler
through the new primary; moved all leader/pager/e2e harness setters to
KIGI_API_KEY; made every house-key isolation test unset KIGI_API_KEY too;
updated user-facing hints to name KIGI_API_KEY.

Tests: e2e proves enrichment-supplied context (wire carries none), non-vacuous
tool_call restriction, bare-id round-trip under xai/, Passthrough; validation
tests hit /models (401 reject, 200 accept); house_env_var_takes_precedence_over_xai
pins the new precedence. Registry at 16; picker 17 rows; 4 auth arrays + xai.

Review (16 findings, all fixed): caught a missed else-branch env clear in the
paste-key handler (would leak the house key past a clear) and a non-hermetic
credential-priority test; both fixed.
This commit is contained in:
ZacharyZhang-NY
2026-07-21 16:23:08 -04:00
parent 193d16f6d5
commit ebf11057f8
25 changed files with 297 additions and 56 deletions
Download Patch File Download Diff File Expand all files Collapse all files
crates/codegen/kigi-shell/src/agent/auth_method.rs
+91 -16
View File
@@ -19,24 +19,36 @@ pub(crate) fn new_shared_auth_method_id(initial: Option<acp::AuthMethodId>) -> S
))
}
/// Env var that, when set, advertises `xai.api_key` as a viable auth method.
/// Primary env var that, when set, advertises `xai.api_key` as a viable auth
/// method. NOTE: `xai.api_key` is the *house* bring-your-own-key method (the
/// upstream product is house-branded "xai"), unrelated to the x.ai/Grok
/// provider. That collision is why the primary env moved here to `KIGI_API_KEY`
/// — `XAI_API_KEY` is now the x.ai/Grok provider key (see `XAI_SPEC`).
///
/// Kept as a constant so test code and the production check stay in sync.
pub const HOUSE_API_KEY_ENV_VAR: &str = "KIGI_API_KEY";
/// Back-compat fallback env: `XAI_API_KEY` was the house BYOK key before it
/// became the x.ai/Grok provider key. Still honored so existing house-BYOK
/// deployments keep working (they share the key with the Grok provider).
pub const XAI_API_KEY_ENV_VAR: &str = "XAI_API_KEY";
/// Legacy env var name. Checked as a fallback when `XAI_API_KEY` is not set,
/// so existing deployments that use the old name keep working.
/// Legacy env var name (pre-`XAI_API_KEY`). Checked last so the oldest
/// deployments keep working.
pub const LEGACY_XAI_API_KEY_ENV_VAR: &str = "KIGI_CODE_XAI_API_KEY";
/// Read the API key from the environment.
/// Read the house BYOK API key from the environment.
///
/// Checks `XAI_API_KEY` first, then falls back to the legacy
/// `KIGI_CODE_XAI_API_KEY` for backward compatibility.
/// Checks `KIGI_API_KEY` first, then the back-compat `XAI_API_KEY`, then the
/// legacy `KIGI_CODE_XAI_API_KEY`.
pub fn read_xai_api_key_env() -> Result<String, std::env::VarError> {
std::env::var(XAI_API_KEY_ENV_VAR).or_else(|_| std::env::var(LEGACY_XAI_API_KEY_ENV_VAR))
std::env::var(HOUSE_API_KEY_ENV_VAR)
.or_else(|_| std::env::var(XAI_API_KEY_ENV_VAR))
.or_else(|_| std::env::var(LEGACY_XAI_API_KEY_ENV_VAR))
}
/// Returns `true` if either `XAI_API_KEY` or `KIGI_CODE_XAI_API_KEY` is set.
/// Returns `true` if any house BYOK env is set: `KIGI_API_KEY` (primary) or the
/// back-compat `XAI_API_KEY` / `KIGI_CODE_XAI_API_KEY`.
pub fn has_xai_api_key_env() -> bool {
read_xai_api_key_env().is_ok()
}
@@ -271,7 +283,7 @@ pub fn session_token_auth_gate(
pub const AUTH_ERROR_SESSION_EXPIRED: &str =
"Session expired. Run `kigi login` to re-authenticate.";
pub const AUTH_ERROR_API_KEY: &str = "Authentication failed. Run `kigi login`, set XAI_API_KEY, or add api_key to ~/.kigi/config.toml.";
pub const AUTH_ERROR_API_KEY: &str = "Authentication failed. Run `kigi login`, set KIGI_API_KEY, or add api_key to ~/.kigi/config.toml.";
/// Next ACP method id when `cached_token` cannot proceed (missing / expired):
/// prefer non-interactive `xai.api_key` when advertiseable, else the
@@ -292,7 +304,7 @@ pub fn xai_api_key_auth_method() -> acp::AuthMethod {
"xai.api_key".to_string(),
)
.description(Some(format!(
"{XAI_API_KEY_ENV_VAR} or api_key/env_key in config.toml"
"{HOUSE_API_KEY_ENV_VAR} or api_key/env_key in config.toml"
))),
)
}
@@ -607,7 +619,8 @@ mod tests {
"together",
"cerebras",
"nvidia",
"vercel-ai-gateway"
"vercel-ai-gateway",
"xai"
]
);
assert_eq!(default_id(&built), Some(XAI_API_KEY_METHOD_ID));
@@ -645,7 +658,8 @@ mod tests {
"together",
"cerebras",
"nvidia",
"vercel-ai-gateway"
"vercel-ai-gateway",
"xai"
]
);
assert_eq!(default_id(&built), Some(CACHED_TOKEN_AUTH_METHOD_ID));
@@ -676,7 +690,8 @@ mod tests {
"together",
"cerebras",
"nvidia",
"vercel-ai-gateway"
"vercel-ai-gateway",
"xai"
]
);
assert_eq!(default_id(&built), Some(CACHED_TOKEN_AUTH_METHOD_ID));
@@ -710,7 +725,8 @@ mod tests {
"together",
"cerebras",
"nvidia",
"vercel-ai-gateway"
"vercel-ai-gateway",
"xai"
]
);
assert_eq!(default_id(&built), None);
@@ -759,21 +775,34 @@ mod tests {
#[test]
#[serial]
fn legacy_env_var_fallback_advertises_xai_api_key() {
let _house = EnvGuard::unset(HOUSE_API_KEY_ENV_VAR);
let _unset = EnvGuard::unset(XAI_API_KEY_ENV_VAR);
let _set = EnvGuard::set(LEGACY_XAI_API_KEY_ENV_VAR, "legacy-key");
assert!(has_xai_api_key_env());
assert_eq!(read_xai_api_key_env().unwrap(), "legacy-key");
}
/// The new env var takes precedence over the legacy one.
/// `XAI_API_KEY` takes precedence over the older legacy env var.
#[test]
#[serial]
fn new_env_var_takes_precedence_over_legacy() {
fn xai_env_var_takes_precedence_over_legacy() {
let _house = EnvGuard::unset(HOUSE_API_KEY_ENV_VAR);
let _new = EnvGuard::set(XAI_API_KEY_ENV_VAR, "new-key");
let _legacy = EnvGuard::set(LEGACY_XAI_API_KEY_ENV_VAR, "legacy-key");
assert_eq!(read_xai_api_key_env().unwrap(), "new-key");
}
/// After the migration, `KIGI_API_KEY` is the house BYOK primary and wins
/// over the back-compat `XAI_API_KEY` (now the x.ai/Grok provider key).
#[test]
#[serial]
fn house_env_var_takes_precedence_over_xai() {
let _house = EnvGuard::set(HOUSE_API_KEY_ENV_VAR, "house-key");
let _xai = EnvGuard::set(XAI_API_KEY_ENV_VAR, "grok-key");
let _legacy = EnvGuard::set(LEGACY_XAI_API_KEY_ENV_VAR, "legacy-key");
assert_eq!(read_xai_api_key_env().unwrap(), "house-key");
}
/// Moonshot authenticate with no configured key: actionable error naming
/// the platform, the login screen, and the platform-scoped env var. No
/// HTTP is attempted (`key: None` short-circuits).
@@ -931,4 +960,50 @@ mod tests {
.await
.expect("200 from /key must validate the key");
}
/// xAI has no validation-path override: `/v1/models` itself requires auth
/// (401 without a valid key), so it doubles as the validator. A bad key is
/// rejected.
#[tokio::test]
#[serial]
async fn xai_validates_against_models_and_rejects_bad_key() {
use wiremock::matchers::{method, path};
let server = wiremock::MockServer::start().await;
wiremock::Mock::given(method("GET"))
.and(path("/models"))
.respond_with(wiremock::ResponseTemplate::new(401))
.expect(1)
.mount(&server)
.await;
let _base = EnvGuard::set(kigi_models::XAI_BASE_URL_ENV, &server.uri());
let err = authenticate_platform_api_key(kigi_models::PlatformId::Xai, Some("xai-bad"))
.await
.expect_err("a 401 from /models must reject the key");
assert_eq!(
err.message,
"Invalid API key for xai \u{2014} check your key on console.x.ai"
);
}
/// A valid xAI key: `/models` returns 200 for the Bearer header → accepted.
#[tokio::test]
#[serial]
async fn xai_valid_key_succeeds_via_models() {
use wiremock::matchers::{header, method, path};
let server = wiremock::MockServer::start().await;
wiremock::Mock::given(method("GET"))
.and(path("/models"))
.and(header("Authorization", "Bearer xai-good"))
.respond_with(
wiremock::ResponseTemplate::new(200)
.set_body_json(serde_json::json!({ "data": [] })),
)
.expect(1)
.mount(&server)
.await;
let _base = EnvGuard::set(kigi_models::XAI_BASE_URL_ENV, &server.uri());
authenticate_platform_api_key(kigi_models::PlatformId::Xai, Some("xai-good"))
.await
.expect("200 from /models must validate the key");
}
}
crates/codegen/kigi-shell/src/agent/config.rs
+13 -2
View File
@@ -4893,7 +4893,9 @@ reasoning_effort = "low"
#[test]
#[serial]
fn resolve_credentials_empty_env_key_falls_through_to_global_key() {
use crate::agent::auth_method::{LEGACY_XAI_API_KEY_ENV_VAR, XAI_API_KEY_ENV_VAR};
use crate::agent::auth_method::{
HOUSE_API_KEY_ENV_VAR, LEGACY_XAI_API_KEY_ENV_VAR, XAI_API_KEY_ENV_VAR,
};
use kigi_chat_state::AuthType;
use kigi_test_support::EnvGuard;
let sentinel = "xai-global-sentinel-key";
@@ -4901,6 +4903,9 @@ reasoning_effort = "low"
let alias = "KIGI_TEST_EMPTY_ENV_GLOBAL_ALIAS";
let _primary = EnvGuard::set(primary, "");
let _alias = EnvGuard::set(alias, "");
// Unset the new house primary so the back-compat XAI_API_KEY is the
// global key under test (not shadowed by an ambient KIGI_API_KEY).
let _house = EnvGuard::unset(HOUSE_API_KEY_ENV_VAR);
let _global = EnvGuard::set(XAI_API_KEY_ENV_VAR, sentinel);
let _legacy = EnvGuard::unset(LEGACY_XAI_API_KEY_ENV_VAR);
let mut model = test_model_entry("m", "https://inference.example/v1", None, None, None);
@@ -6454,7 +6459,13 @@ reasoning_effort = "low"
None,
None,
);
unsafe { std::env::set_var("XAI_API_KEY", "env-key") };
// Clear the house primary + legacy so the XAI_API_KEY set below is the
// env key resolved by read_xai_api_key_env (KIGI_API_KEY wins otherwise).
unsafe {
std::env::remove_var("KIGI_API_KEY");
std::env::remove_var("KIGI_CODE_XAI_API_KEY");
std::env::set_var("XAI_API_KEY", "env-key");
}
let sampling = resolve_sampling(&model_with_key, Some("session-key"));
assert_eq!(
sampling.api_key.as_deref(),
crates/codegen/kigi-shell/src/agent/models_fetch.rs
+91 -1
View File
@@ -132,7 +132,7 @@ fn fetch_custom_endpoint_models_blocking(
.ok_or(std::env::VarError::NotPresent)
})
.map_err(|_| {
BackendError::Auth("No API key for custom models endpoint. Set XAI_API_KEY.".into())
BackendError::Auth("No API key for custom models endpoint. Set KIGI_API_KEY.".into())
})?;
let request = client
.get(&url)
@@ -1994,6 +1994,96 @@ mod tests {
);
}
/// xAI-cycle e2e: /v1/models is minimal (bare grok ids, no context), so
/// enrichment supplies context/limits; the restriction drops the
/// non-tool-calling grok-imagine-* generators; bare id round-trips under
/// the `xai/` key; Passthrough dialect.
#[tokio::test(flavor = "multi_thread")]
#[serial_test::serial]
async fn xai_enriches_restricts_and_maps_passthrough() {
let platform_server = wiremock::MockServer::start().await;
wiremock::Mock::given(wiremock::matchers::method("GET"))
.and(wiremock::matchers::path("/models"))
.respond_with(wiremock::ResponseTemplate::new(200).set_body_json(
// xAI /v1/models is minimal: id/object only, NO context field —
// so a non-zero context below can only come from enrichment.
serde_json::json!({ "data": [
{ "id": "grok-4.5", "object": "model" },
{ "id": "grok-imagine-image", "object": "model" }
]}),
))
.expect(1)
.mount(&platform_server)
.await;
let modelsdev_server = wiremock::MockServer::start().await;
wiremock::Mock::given(wiremock::matchers::method("GET"))
.and(wiremock::matchers::path("/api.json"))
.respond_with(wiremock::ResponseTemplate::new(200).set_body_json(
serde_json::json!({ "xai": { "models": {
"grok-4.5": {
"limit": {"context": 500000, "output": 128000},
"tool_call": true
},
// present in enrichment too, but not tool-calling → dropped.
"grok-imagine-image": { "limit": {"context": 8000} }
}}}),
))
.expect(1)
.mount(&modelsdev_server)
.await;
let cache_dir = tempfile::tempdir().unwrap();
let _base =
kigi_test_support::EnvGuard::set(kigi_models::XAI_BASE_URL_ENV, platform_server.uri());
let _mdev = kigi_test_support::EnvGuard::set(
crate::agent::enrichment_fetch::MODELS_DEV_URL_ENV,
format!("{}/api.json", modelsdev_server.uri()),
);
let _mdev_cache = kigi_test_support::EnvGuard::set(
crate::agent::enrichment_fetch::MODELS_DEV_CACHE_DIR_ENV,
cache_dir.path(),
);
let endpoints = crate::agent::config::EndpointsConfig::default();
let keys = crate::agent::models::PlatformApiKeys::test_single(
kigi_models::PlatformId::Xai,
"xai-1",
);
let result = tokio::task::spawn_blocking(move || {
fetch_platform_models_blocking(&endpoints, None, &keys)
})
.await
.unwrap()
.expect("fetch must succeed");
assert_eq!(
result
.models
.iter()
.map(|m| m.id.as_deref().unwrap_or_default())
.collect::<Vec<_>>(),
vec!["xai/grok-4.5"],
"grok-imagine-image (not tool-calling) dropped; bare id kept under the xai key"
);
let entry = &result.models[0];
assert_eq!(
entry.context_window.get(),
500_000,
"context comes from enrichment (the wire listing carries none)"
);
assert_eq!(entry.max_completion_tokens, Some(128_000));
assert_eq!(entry.model, "grok-4.5", "the bare grok id rides the wire");
let model_entry = crate::agent::config::ModelEntry::from_config_entry(entry);
let creds = crate::agent::config::ResolvedCredentials {
api_key: Some("xai-1".into()),
base_url: entry.base_url.clone(),
auth_type: kigi_chat_state::AuthType::ApiKey,
auth_scheme: Default::default(),
};
let cfg = crate::agent::config::sampling_config_for_model(&model_entry, creds, None);
assert_eq!(
cfg.chat_compat,
kigi_sampling_types::ChatCompat::Passthrough
);
}
#[test]
fn get_env_keys_parses_strings_and_rejects_non_strings() {
use crate::agent::config::EnvKeys;
crates/codegen/kigi-shell/src/agent/mvp_agent/acp_agent.rs
+2 -2
View File
@@ -165,7 +165,7 @@ impl acp::Agent for MvpAgent {
&crate::util::kigi_home::kigi_home(),
)
{
unsafe { std::env::set_var("XAI_API_KEY", &api_key) };
unsafe { std::env::set_var("KIGI_API_KEY", &api_key) };
tracing::info!("auth: loaded API key from auth.json (xai::api_key scope)");
kigi_log::unified_log::info(
"auth: loaded API key from auth.json (xai::api_key scope)",
@@ -357,7 +357,7 @@ impl acp::Agent for MvpAgent {
return Err(
acp::Error::auth_required()
.data(
"Set XAI_API_KEY or add api_key/env_key to config.toml.",
"Set KIGI_API_KEY or add api_key/env_key to config.toml.",
),
);
}
crates/codegen/kigi-shell/src/agent/mvp_agent/tests.rs
+3 -1
View File
@@ -1904,10 +1904,12 @@ async fn cached_token_fallthrough_prefers_api_key_for_deployment_key() {
#[serial_test::serial]
async fn cached_token_fallthrough_falls_to_kigi_com_without_credentials() {
use crate::agent::auth_method::{
KIMI_CODE_METHOD_ID, LEGACY_XAI_API_KEY_ENV_VAR, XAI_API_KEY_ENV_VAR,
HOUSE_API_KEY_ENV_VAR, KIMI_CODE_METHOD_ID, LEGACY_XAI_API_KEY_ENV_VAR,
XAI_API_KEY_ENV_VAR,
};
use kigi_test_support::EnvGuard;
let _lockdown = EnvGuard::unset("KIGI_DISABLE_API_KEY_AUTH");
let _house = EnvGuard::unset(HOUSE_API_KEY_ENV_VAR);
let _new = EnvGuard::unset(XAI_API_KEY_ENV_VAR);
let _legacy = EnvGuard::unset(LEGACY_XAI_API_KEY_ENV_VAR);
let agent = build_minimal_agent_for_tests();