José Almeida
2779fb24aa
Lay the backend groundwork so the existing mock-agent can run inside local_docker containers — needed for real-Docker E2E tests that don't require Anthropic/OpenAI API keys. Container builder bind-mounts a host mock-agent binary at /usr/local/bin/mock-agent when KANDEV_MOCK_AGENT_LINUX_BINARY is set (or build/mock-agent-linux-amd64 sits next to the kandev binary), the same pattern already used for agentctl. MockAgent.BuildCommand always resolves via PATH so the same command works on host (E2E PATH includes bin/) and in containers (mounted at /usr/local/bin/mock-agent). Mock now declares non-nil empty RemoteAuth so it counts as a remote-capable agent. The remoteauth catalog includes specs even with zero methods, and the frontend gating helper treats an empty-method spec as "no credentials needed" — letting mock-agent pass the configure-on-executor check that hides uncatalogued agents on remote executors. repositoryCloneURL falls back to the local origin remote URL when the repo has no provider info, so local-only repos with a real remote (including file:// remotes used in E2E) work on remote executors that clone inside the container. Makefile: build-mock-agent-linux target mirrors build-agentctl-linux.
53 lines
2.1 KiB
TypeScript
53 lines
2.1 KiB
TypeScript
import type { ExecutorProfile } from "@/lib/types/http";
|
|
import type { RemoteAuthSpec } from "@/lib/api/domains/settings-api";
|
|
|
|
const REMOTE_EXECUTOR_TYPES = new Set(["local_docker", "remote_docker", "sprites"]);
|
|
|
|
export function executorRequiresAgentCredentials(executorType?: string | null): boolean {
|
|
if (!executorType) return false;
|
|
return REMOTE_EXECUTOR_TYPES.has(executorType);
|
|
}
|
|
|
|
function parseJSON<T>(raw: string | undefined, fallback: T): T {
|
|
if (!raw) return fallback;
|
|
try {
|
|
return JSON.parse(raw) as T;
|
|
} catch {
|
|
return fallback;
|
|
}
|
|
}
|
|
|
|
/**
|
|
* For local/worktree executors, agents need no extra credentials → always supported.
|
|
* For remote executors (Docker/Sprites), the executor profile must carry either:
|
|
* - a non-env auth method ID for the agent's spec in `remote_credentials`, or
|
|
* - a non-null secret keyed by an env method ID in `remote_auth_secrets`.
|
|
* Agents without a remote-auth spec (Copilot/Amp/OpenCode/TUI) cannot
|
|
* carry credentials on remote executors → blocked. A spec with zero methods
|
|
* means "no credentials needed" (mock-agent for tests) → allowed.
|
|
*
|
|
* Spec IDs are registry-type strings ("claude-acp", "codex-acp", …) which the
|
|
* frontend exposes as `AgentProfileOption.agent_name`. `agent_id` is a DB UUID
|
|
* and is unrelated to the catalog.
|
|
*/
|
|
export function isAgentConfiguredOnExecutor(
|
|
agent: { agent_name: string },
|
|
executorProfile: Pick<ExecutorProfile, "config" | "executor_type">,
|
|
authSpecs: RemoteAuthSpec[],
|
|
): boolean {
|
|
if (!executorRequiresAgentCredentials(executorProfile.executor_type)) return true;
|
|
|
|
const spec = authSpecs.find((s) => s.id === agent.agent_name);
|
|
if (!spec) return false;
|
|
if (spec.methods.length === 0) return true;
|
|
|
|
const credentials = new Set(parseJSON<string[]>(executorProfile.config?.remote_credentials, []));
|
|
if (spec.methods.some((m) => m.type !== "env" && credentials.has(m.method_id))) return true;
|
|
|
|
const secrets = parseJSON<Record<string, string | null>>(
|
|
executorProfile.config?.remote_auth_secrets,
|
|
{},
|
|
);
|
|
return spec.methods.some((m) => m.type === "env" && secrets[m.method_id]);
|
|
}
|