continuedev/continue · error · Error

Depth must be a number

Error message

Depth must be a number

What it means

The WatsonX streaming chat request returned a non-ok status or had no body, and the raw response text is embedded. This is the generic upstream failure for WatsonX chat streaming (auth, space/project id, model id, network).

Source

Thrown at core/context/providers/GitCommitContextProvider.ts:111

            name: query,
            description: `commit ${query}`,
            content,
          },
        ];
      }
    } catch (err) {
      return [];
    }
  }

  async loadSubmenuItems(
    args: LoadSubmenuItemsArgs,
  ): Promise<ContextSubmenuItem[]> {
    const depth = this.options?.Depth ?? 50;
    const lastXCommitsDepth = this.options?.LastXCommitsDepth ?? 10;

    if (typeof depth !== "number") {
      throw new Error("Depth must be a number");
    }
    if (typeof lastXCommitsDepth !== "number") {
      throw new Error("LastXCommitsDepth must be a number");
    }

    const topLevelDir = fileURLToPath((await args.ide.getWorkspaceDirs())[0]);
    try {
      const gitResult = await runGitCommand(
        [
          "--no-pager",
          "log",
          '--pretty=format:"%H%x00%s"',
          `-n`,
          depth.toString(),
        ],
        topLevelDir,
      );
      const recentCommits = [

View on GitHub (pinned to 5522c6f44c)

Solutions

1. Read the embedded response.text() — WatsonX states the exact reason (401 token, 404 project/model)
2. Refresh the IAM/access token before long-lived streaming sessions
3. Verify project/space id and that the model is available in your WatsonX region
4. Wrap stream start in retry-with-backoff for transient 5xx

try { for await (const c of api.chatCompletionStream(body, signal)) {} } catch (e) { const msg = (e as Error).message; if (/401|token/i.test(msg)) { await refreshToken(); return retry(); } if (/429|5\d\d/.test(msg)) await backoff(); else throw e; }

When it happens

Trigger: Expired/invalid IAM token in headers; wrong project_id or watsonx.ai space; model id not available to the account; TLS/proxy issues producing non-2xx; empty body on early connection close.

Common situations: IAM token from WatsonX Machine Learning expires mid-session and streams start failing; wrong WATSONX_PROJECT_ID env var; model gated by entitlement in the WatsonX catalog.

Related errors

• GitLab Private Token is required!
• Error getting prompt: MCP Connection ${serverName} not found
• Unsupported transport type: ${this.options.type}
• MCP Connection ${serverId} not found
• no query params found

AI-assisted analysis of continuedev/continue@5522c6f44c (2026-08-27). Data as JSON: /api/errors/65acce773f6d13bd. Report an issue: GitHub.