Fixes for Claude 5 Prompt Generation and Quota Alerts

Version: 2.4.4, Date: July 28, 2026

Prompt generation was failing outright on Claude 5 and other reasoning models, and tracing that failure turned up a cluster of related problems in how quotas were enforced and how limit messages were worded. This release fixes the generation failure and cleans up the quota surface around it, so the messages you see match the limit you actually hit.

Templates

• Fixed: Prompt generation returned a “Failed to generate executable prompt” error whenever a Claude 5 or other reasoning model was selected. Reasoning models return their response in a different shape and only part of it was being read, so the result came back empty even though the model had done the work. Full responses are now read correctly and prompts generate normally. This first showed up in the YouTube template but affected every feature that could run one of those models.

Quota Usage

• Fixed: Limits on Prompt Cards and Section Libraries were silently not being applied, so both behaved as unlimited regardless of your plan. Both now enforce the allowance your plan actually defines.
• Fixed: Credit allotments on lifetime and AppSumo plans are now resolved from the correct plan type, so the credits you see and spend match what your plan grants.
• Fixed: For anyone using their own API key, the first run of each month failed to record against your usage counters. Own-key usage now records cleanly from the first run onward.
• Improved: Per-request ceilings that your plan defines are now applied before a run starts, so an oversized request is caught up front with a clear message rather than partway through. These apply to platform credit runs only: usage on your own API key is never capped by us.

Workflows

• Fixed: When a workflow pauses partway through a run because a limit was reached, the pause now carries the correct limit information through to the message you see, instead of a generic one.

Across the App

• Fixed: Quota messages are now typed correctly everywhere they appear. Previously every limit message said you were out of credits, whatever the actual cause. Analyzers, workflows, templates, custom prompt cards, URL to Prompt Kit, and the intent builder now all report the specific limit you reached.
• Improved: Limit messages respect the difference between platform runs and own-key runs. Platform runs are always described in credits, and own-key runs report token usage, so the number you are shown is the one that applies to you.
• Fixed: The credit reset date shown in a limit message no longer disagrees with your Quota Usage and subscription pages. All three now read from the same source and show the same date.