Add proper binary guards for authorize_refresh/1

[michaeljguarino]

Trying to plug random conditions that htrow errors.

Test Plan

Test environment: https://console.your-env.onplural.sh/

Checklist

• I have added a meaningful title and summary to convey the impact of this PR to a user.
• If required, I have updated the Plural documentation accordingly.
• I have added tests to cover my changes.
• I have deployed the agent to a test environment and verified that it works as expected (required only when changing agent code).

Plural Flow: console

[soffi-ai]

Soffi AI Summary

This PR addresses two distinct stability and behavioral improvements in the chatbot/AI subsystem:

1. Binary guard for authorize_refresh/1 — Adds a proper when is_binary(token) guard to the function and a fallback clause returning {:error, "no refresh token provided"}. This prevents runtime errors when nil or non-binary values are passed as refresh tokens, hardening the auth path against unexpected input.

2. Improved chatbot reply behavior — Changes how follow-up messages (replies to prior chatbot messages in a thread) are handled. Previously, a reply would create a WorkbenchJobActivity directly and mark the job as :pending, which could conflict with an already-running job. Now, replies create a QueuedPrompt with an immediate dequeable_at timestamp, allowing the system to enqueue the follow-up safely regardless of the job's current status (running, successful, etc.). A dedicated EEx prompt template (chat_reply.md.eex) is introduced to render the follow-up context separately from new-message prompts, giving the agent clear instructions to continue the existing job thread. The parent_job/1 helper is also simplified to return the job regardless of its status, enabling replies even when the parent job is still active.

Commits

Commit	Summary
39c25ca	Adds a when is_binary(token) guard to authorize_refresh/1 and a fallback clause to return a clean error tuple instead of crashing when called with a non-binary (e.g., nil) value.
52952a8	Switches chatbot thread replies from creating a WorkbenchJobActivity to creating a QueuedPrompt with an immediate dequeue time, enabling safe reply enqueueing regardless of the parent job's current status. Simplifies parent_job/1 to return the job unconditionally and adds a new chat_reply.md.eex prompt template for follow-up context.
7315c57	Tunes the chatbot reply prompt and related test assertions, refining the reply message wording and verifying the new QueuedPrompt fields and job state expectations.

Deploy in Soffi

---

Updated: 2026-07-28 23:08 UTC

[greptile-apps]

Greptile Summary

Adds defensive refresh-token handling.

• Restricts the existing refresh-token lookup path to binary tokens.
• Returns an explicit error tuple when authorize_refresh/1 receives any non-binary value.

Confidence Score: 5/5

The PR appears safe to merge, with no actionable defects identified in the changed refresh-token handling.

The normal GraphQL path already supplies a required string token, while direct non-binary calls now return the service's established error-tuple shape instead of entering the database lookup.

Important Files Changed

Filename	Overview
lib/console/services/users.ex	Adds a binary guard and safe fallback to prevent malformed refresh-token values from reaching the token lookup.

_{Reviews (1): Last reviewed commit: "Add proper binary guards for authorize_r..." | Re-trigger Greptile}

[michaeljguarino]

plural deploy this

[plural-copilot]

Deploy PR created: https://github.com/pluralsh/plrl-up-demos/pull/2397

[michaeljguarino]

plural deploy this

[plural-copilot]

Created deploy PR: https://github.com/pluralsh/plrl-up-demos/pull/2398

[michaeljguarino]

plural deploy this

[plural-copilot]

Deployment PR created: https://github.com/pluralsh/plrl-up-demos/pull/2400