Problem
With both first-party blueprints and community-submitted WASM parsers able to produce a translated sentence, the dashboard currently has no way (per the README) to tell a user which mechanism produced a given sentence, or whether that parser is sandboxed/untrusted vs. an official blueprint. For a transparency tool, hiding the provenance of the translation itself is an odd blind spot.
Proposed work
Add a small badge/tooltip on each translated event indicating its source: Official blueprint, Community WASM parser (sandboxed), or Unrecognized — fallback.
For WASM-sourced translations, surface the execution result metadata already available from WasmSandboxRunner (e.g. execution time, success flag) in an expandable detail panel.
Link from the badge to the relevant blueprint/parser source file or registry entry where possible.
Acceptance criteria
All three provenance states render distinctly and correctly using realistic mock data for each case.
WASM-sourced events show real sandbox execution metadata, not placeholder text.
No UI regression for events that have no provenance metadata (e.g. very old cached events).
Problem
With both first-party blueprints and community-submitted WASM parsers able to produce a translated sentence, the dashboard currently has no way (per the README) to tell a user which mechanism produced a given sentence, or whether that parser is sandboxed/untrusted vs. an official blueprint. For a transparency tool, hiding the provenance of the translation itself is an odd blind spot.
Proposed work
Add a small badge/tooltip on each translated event indicating its source: Official blueprint, Community WASM parser (sandboxed), or Unrecognized — fallback.
For WASM-sourced translations, surface the execution result metadata already available from WasmSandboxRunner (e.g. execution time, success flag) in an expandable detail panel.
Link from the badge to the relevant blueprint/parser source file or registry entry where possible.
Acceptance criteria
All three provenance states render distinctly and correctly using realistic mock data for each case.
WASM-sourced events show real sandbox execution metadata, not placeholder text.
No UI regression for events that have no provenance metadata (e.g. very old cached events).