Skip to content

llm-council: 429/ETIMEDOUT errors produce unhelpful messages — no actionable hint about cause #100

Description

@23r2efewvcs

Enhancement

When a model call fails with 429 (rate-limit) or ETIMEDOUT (connection timeout), the error message is generic:

[ERROR 429: {"error":{"message":"Provider returned error","code":429,...}}]

or for connection errors, the raw Node.js error propagates and kills the session.

Users have no idea:

  • Why it failed (another client using the same API key? free-tier concurrent limit?)
  • What to do (wait? switch provider? use --sequential? use a separate key?)

Current behavior

if (res.status >= 400) return { success: false, content: `[ERROR ${res.status}: ${res.body.slice(0, 300)}]`, model, latency_ms: elapsed };

Generic error. No context about what the status code means.

Proposed fix

After retries are exhausted, produce differentiated error messages:

  • 429: explain it's a rate-limit, suggest: wait 60s, use --sequential, switch provider, use separate API key
  • ETIMEDOUT/ECONNRESET: explain connection failed, suggest: increase --timeout, use --sequential, try different endpoint
  • 5xx: explain server error, suggest: retry later, check provider status

All messages are [ERROR ...] with a \nHint: section.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions