Codex
d04baf5c1d
3.0 KiB
Priorities
The ranked work queue for the autonomous improvement loop. The architecture-review pass (the architect) owns this file: each run it turns the roadmap plus an internal scan (gaps in the services → agents → workflows lifecycle, API coherence, drift, tech debt, test and DX friction) into a single ordered list — highest-value first — and links each item to a tracking issue. The hourly continuous-improvement pass works the top item whose issue is still open. So the architect decides what, and the increment loop builds it.
Reading / editing. An item is done when its linked issue closes (the increment
that builds it adds Closes #<issue>). Roadmap phase (Now → Next → Later) is the
primary ordering; internal findings are interleaved by value, not kept in a
separate list. The human can reorder this list — or the issues — at any time to
redirect the loop; direction always wins.
Off-limits to the loop (the architect proposes these as notes, never as queue items the loop can auto-merge): brand/positioning copy, breaking public-API changes, architectural rewrites. Those go to the human.
Work queue (ranked)
- Add agent model retry/backoff contract (#4500) — With the AtlasCloud marker gap and first-agent docs command parity now shipped, the highest-value remaining Now-phase hardening gap is provider-free coverage for transient model failures, retry budgets, backoff, and cancellation in the agent loop. This keeps cross-provider reliability moving without requiring live credentials or broad public-API changes.
- Add first-agent docs wayfinding link contract (#4508) — The adoption story is now stronger, but the on-ramp spans README, website guides, CLI command output, examples, debugging, and 0→hero references. Add a provider-free check that fails on stale links or missing required wayfinding steps so the first-agent path remains walkable as docs and examples evolve.
- Stabilize AtlasCloud agent-flow onboarding side effects (#4503) — The live AtlasCloud harness exposed a services → agents → workflows seam where a timeout can replay workspace creation without completing notification. Exact-once side-effect behavior is core to operable workflows, but it ranks after the provider-free retry contract and one adoption guardrail because it is live-provider-specific.
- Make AtlasCloud universe A2A reachability probe deterministic (#4504) — The universe checkout flow completed, but the A2A reachability probe timed out under AtlasCloud. This matters for cross-framework operability and agent discoverability, yet it is narrower than the retry/idempotency issues because the durable workflow side effects already passed.
Seeded by Claude Code from the roadmap + open issues; thereafter maintained by the architecture-review pass.