feat(connectors): add Ollama and LM Studio as a local auth kind

Both servers expose OpenAI-compatible /v1/chat/completions and /v1/models,
so we route them through pi-ai's `openai-completions` API the same way
Fireworks does — but with a per-installation base URL the user enters
during onboarding instead of a hardcoded one. The shared schemas grow a
`local` credential payload carrying just the baseUrl; the credential
resolver surfaces it on `ResolvedCredential.baseUrl`, the runtime bridge
forwards it through `prepareRun`, and the worker plumbs it into
`resolveModel` so the streamed Model<openai-completions> targets the
user's server. The local-cost path short-circuits to zero in cost-writer
since users bill their own hardware, and the live model list is fetched
from the saved baseUrl by `fetchLocalConnectorModels` (Ollama and LM
Studio both speak the standard OpenAI catalog shape). Onboarding ships
a new LocalProviderCard and the Settings panel a LocalForm; both default
to the catalog's per-connector localhost URL and accept any custom URL.
This commit is contained in:
ZacharyZhang-NY committed 2026-05-14 14:55:38 -04:00
1 parent 07d9c00fab
commit a532120649
26 files changed
+504 -18

No files matched your search

+3 -1
View File
@@ -107,12 +107,14 @@ describe("pi-mono migrations 27 + 28", () => {
db.init();
const rows = db.projection.snapshot.listSnapshot().connectors.map((entry) => entry.id);
expect(rows.length).toBe(26);
expect(rows.length).toBe(28);
expect(rows).toContain("claude-pro");
expect(rows).toContain("amazon-bedrock");
expect(rows).toContain("google-vertex");
expect(rows).toContain("fireworks");
expect(rows).toContain("fireworks-firepass");
expect(rows).toContain("ollama");
expect(rows).toContain("lm-studio");
expect(rows.find((id) => (id as string) === "claude_local")).toBeUndefined();
expect(rows.find((id) => (id as string) === "codex_local")).toBeUndefined();
expect(rows.find((id) => (id as string) === "claude_compatible")).toBeUndefined();