Releases: dannyvfilms/gptemailsummary
V4.1
Shortcut (ElevenLabs): https://www.icloud.com/shortcuts/820a1be086d444168fa14312230aa093
Shortcut (Siri): https://www.icloud.com/shortcuts/f87f7b4839064904896ce642f068581f
Shortcut changes:
- No changes.
Script Changes:
- OpenAI Prompt adjusts if no emails are present, ensuring it does not fabricate emails to summarize.
- OpenAI Prompt should adjust if code errors are caught in the function, and include the errors in the prompt.
- Locks will prevent a new payload from being generated if a summary is currently being made (to prevent time-out errors).
- Lock was supposed to interrupt payload generation if a summary was being made, but that may need to be revisited.
V4
Updated Shortcut (ElevenLabs): https://www.icloud.com/shortcuts/820a1be086d444168fa14312230aa093
Updated Shortcut (Siri): https://www.icloud.com/shortcuts/f87f7b4839064904896ce642f068581f
Shortcut changes:
- Removed Fetch Latest Emails command in favor of constantly generating payload server-side
- Removes automatic Mark Emails as Read function in favor of button (useful in case of Request Timeouts)
- Changed Statistics to use Quick Look
Script Changes:
- Rewrote the entire script to use the IMAP Protocol instead of Gmail's API
- Email payload is recreated every minute to reduce wait time for summary (use gpt3.5-turbo for faster results than gpt-4)
- Multi-Account Support is now available
- Mark Emails as Read is now a manual action
- <container_ip>:1337 will now load a draft of the WebUI. No data connections have been made.
- Additional print statements for logging.
Full Changelog: V3...V4
V3
Updated Shortcut (ElevenLabs): https://www.icloud.com/shortcuts/7deb48573ab64f13a666fa8c63373c88
Updated Shortcut (Siri): https://www.icloud.com/shortcuts/563377e1888a481fb2397a5fb3e260cc
Shortcut changes:
- Correctly grabs Summary so "Summary" and "\n" aren't read out loud
- Adds Statistics value for more information after the run finishes
Script Changes:
- Variable Quantity = true/false is now functional
- EMAIL_MAXEMAILS will be used when Variable Quantity = false
- Added additional html code filtering
- Mark emails as read in a separate thread after sending response payload to reduce request timeout instances
Full Changelog: V2.1...V3
V2.1
Updated Shortcut: https://www.icloud.com/shortcuts/3f898be2dc6d4c7386d395bf225f1b2c
- Added additional filtering to
remove_html_and_links
- Included sender and subject to
count_characters
to fix a character limit bypass - Updated regex in Shortcut due to a formatting change
V2
Uses a character limit to dynamically pull more emails to fit in OpenAI API Payload. V3 will use EMAIL_VARIABLEQUANTITY=true
for this workflow, while EMAIL_VARIABLEQUANTITY=false
will rely on EMAIL_MAXEMAILS
for the amount to pull.
Terminal logs are far more verbose, so I added boxes
and a nice summary at the end.
Full Changelog: alpha...V2
V1
Initial release. Pulls 10 emails for summary. View Readme for installation instructions.