@lincyaw/workbench@0.2.0-rc.25
Presmium's workbench panels in React (the documents, a document edited on its rendered page with comments and version history) with the workspace service's client and the preview protocol's workbench side; @lincyaw/dsh places them in presmium studio (expe
AI Security Review
scanned 9h ago · by lpm-firewall-aiThis is warning-level runtime telemetry: host gating and same-origin delivery reduce risk, but collected records can include request text and workspace-specific paths or identifiers. The transcription adoption code itself uses explicit review and apply actions.
Decision evidence
public snapshot- Workbench construction starts operation telemetry automatically; collection proceeds when the host reports it is enabled.
- After an agent request is sent, its composed text is also recorded as telemetry.
- Operation records include the active document identifier and can include changed file paths.
- Telemetry is posted to the same-origin api/presmium/telemetry endpoint.
- The manifest declares no lifecycle scripts, and telemetry is gated by the host's enabled setting.
- The transcription adoption flow requires the user to review and apply a plan; no unrelated network destination was identified.
Source & flagged code
8 flaggedA bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review.
src/places.ts#virtual:normalized:round1View on unpkgThis package version adds a dangerous source file absent from the previous stored version; route for source-aware review.
src/panels/TranscriptionAdoption.tsxView on unpkgFindings
Affected versions and remediation
This report applies to @lincyaw/workbench@0.2.0-rc.25.
See all versions of @lincyaw/workbench for other recorded verdicts.
- Review the evidence and your use of @lincyaw/workbench@0.2.0-rc.25 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.