Skip to content

Latest commit

 

History

History
19 lines (16 loc) · 2.31 KB

File metadata and controls

19 lines (16 loc) · 2.31 KB

Non-Goals

These items are intentionally out of scope for the whiteboard package. They can be added by downstream adopters, but they should not be treated as missing P0 work.

Non-goal Reason
Package-manager distribution The framework should stay copyable, inspectable, and easy to adapt while the rule contract is still early.
Python or Node runtime ports PowerShell and Bash cover the current router/runtime reference surfaces. Python diagnostics and tests are allowed; a third full router implementation would increase drift until adoption contracts stabilize.
Community boilerplate such as contribution templates or stale automation Useful for mature projects, but not needed to validate the harness chain itself.
Monitoring dashboards The current package is a local gate and memory-lane framework, not an operations monitoring system.
Full memory backend, vector store, or shared memory server The core defines routing, source boundaries, lane isolation, and source-preserving memory capsules. Downstream adopters may add explicitly approved operational tooling, but database-centered semantic memory is not part of the default whiteboard core.
Automatic cross-agent memory sharing Shared recall is useful only after lane IDs, source-monitoring fields, and explicit cross-lane rules are in place. Default sharing would weaken the project-isolation goal.
Large compatibility database The compatibility manifest should describe the adapter at hand, not become a global matrix of every agent runtime and version.
Every-turn compatibility scanning Compatibility checks are for setup, client updates, hook or wrapper edits, failed smoke tests, or explicit audits. Running them on ordinary tasks would waste context and latency.
Promotion material or repository-topic tuning Discovery work should not drive core architecture changes.
Full test matrix The public core keeps only a lightweight smoke workflow. Downstream adopters should add deeper tests for their runtime.
Full comparison table against other frameworks Short positioning is enough here. Large comparisons become stale and invite unsupported claims.

Before adding any of these, define the owner, maintenance cost, validation path, and whether the addition belongs in the public core or an adopter-specific fork.