Simple resumption of tasks on startup from long-term memory

[jondwillis]

Background

Every new run of auto-gpt results in an empty memory line "This reminds you of these events from your past:\n\n\n" In order to better resume goals based on previous long term memories, I was having to do a lot of error-prone goal-prompt engineering. This makes the agent more self-aware if it has long-term memory, but also less deterministic.

Related pull-request: #2530
This is an alternative approach to the above PR. If one like that or similar is merged, this can be closed.

Changes

Agents will now have relevant memories based on their agent settings file prompt and user input on startup.

Documentation

Inline in the code.

Test Plan

Existing tests pass.

Risks

• If Auto-GPT has done something Mal-adaptive, or the ai settings files change, and incorrectly recalls memory from previous runs, it could "poison" present and future runs. It is not always easy / not all users know how to clear memory.
• This may break certain existing use-cases or change prompt output unexpectedly for existing users.

Risk Mitigation

• Add --wipe-memory and/or env var (like WIPE_REDIS_ON_START). Or, add something like MEMORY_RESTART_RECALL_MODE=prompt|file_history|None (file_history could be from this PR FEAT: Add Chat History File Functionality for Efficient Auto-GPT Resumption #2530)
• Add documentation about whatever path above is chosen

PR Quality Checklist

• My pull request is atomic and focuses on a single change.
• I have thoroughly tested my changes with multiple different prompts.
• I have considered potential risks and mitigations for my changes.
• I have documented my changes clearly and comprehensively.
• I have not snuck in any "extra" small tweaks changes

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
docs	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Apr 26, 2023 4:43pm

[codecov]

Codecov Report

Patch coverage has no change and project coverage change: -7.37 ⚠️

Comparison is base (0ff471a) 50.76% compared to head (4a00898) 43.39%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2602      +/-   ##
==========================================
- Coverage   50.76%   43.39%   -7.37%     
==========================================
  Files          65       65              
  Lines        2935     2938       +3     
  Branches      492      493       +1     
==========================================
- Hits         1490     1275     -215     
- Misses       1326     1593     +267     
+ Partials      119       70      -49     

Impacted Files	Coverage Δ
autogpt/chat.py	21.95% <0.00%> (-42.61%)	⬇️

... and 14 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[github-actions]

This pull request has conflicts with the base branch, please resolve those so we can evaluate the pull request.

[ntindle]

Can you cover this with a bit of testing

[p-i-]

This is a mass message from the AutoGPT core team.
Our apologies for the ongoing delay in processing PRs.
This is because we are re-architecting the AutoGPT core!

For more details (and for infor on joining our Discord), please refer to:
https://github.com/Significant-Gravitas/Auto-GPT/wiki/Architecting

[Pwuts]

Thanks for submitting! This is an interesting and seemingly useful addition. At the moment the long-term memory is not in use: it was disabled and mostly removed/reworked (#4208) for lack of performance. I have labeled this with fridge so we can find it back and process it when the dust settles.