eandualem / eandualem/agent-backbone
Future review: align features and background jobs with agent collaboration
- Dominant language
- Python
- Stars
- 5
- Forks
- 1
- Avg merge
- 38m
- Merged PRs (30d)
- 89
Description
[from:agent-backbone]
## Purpose
Review existing Backbone features against its core purpose: specialist agents communicating and collaborating toward a shared goal. The control plane should make that collaboration reliable while agents decide how to do the work.
## Future review scope
- Inventory existing features, background jobs and recurring checks; explain the user need and architectural responsibility of each.
- Review scheduler-driven behavior in particular. Identify mechanical workflows that duplicate decisions or follow-up better handled by collaborating agents, and propose removing or simplifying them.
- Distinguish essential delivery, state, audit and recovery infrastructure from product logic that prescribes what agents do. Justify retained mechanisms rather than assuming every existing job belongs.
- Use audit logs and observed failures to prioritize concrete reliability and usability improvements once the core works, instead of continuing to expand the feature set.
- Review consistency with runtime independence, discovery without mandatory agent declarations, ordinary peer agents and the existing safety boundaries.
## Expected outcome
A concise keep/simplify/remove assessment grounded in actual behavior, followed by an agreed cleanup plan. Any later changes should reduce unnecessary machinery while preserving dependable collaboration, delivery evidence and runtime sandboxes.
## Timing
Future review only. Recorded at the owner's request on 2026-09-12; do not begin implementation or remove existing jobs now. Scheduling features are not a requested product direction. Revisit when the owner selects this work.
Contributor guide
Research direction
When the owner selects this review, inventory the existing features, background jobs, recurring checks, audit logs, and observed failures described in the issue. Produce a concise keep/simplify/remove assessment and an agreed cleanup plan, preserving dependable collaboration, delivery evidence, and runtime sandboxes; do not implement changes before selection.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- ai, devtools
- Issue type
- Refactor
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100