You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, we may expose secrets in the run results in the output dataclip. We will be implementing a rule that limits what the runtime writes to the run_result, but would like an additional failsafe of scrubbing this before displaying it to the user on the presentation layer:
just finished the call with Mtuchi and Joe around secrets getting stored in run_result dataclips. We think that (if the additional overhead isn’t too high) it makes sense to scrub dataclips during presentation (i.e., after the liveview gets the JSON from the db, after it converts it to a string, and finally right before it renders it in the code-viewers on run_viewer, attempts_live, and the dataclip viewer.
In addition to this, we’ll handle the actual object in the runtime, but it’s a very good failsafe to scrub the strings right before they’re sent to the HTML view. What do you think? (Roger, is this a terrible idea from a performance perspective? Could we hook up the existing scrubber (using the secrets list from an associated attempt) to scrub dataclips with type run_result once they’re converted to text and are about to be viewed as HTML?)
Notes:
Scrubbing should occur after the liveview receives the JSON from the database.
Scrubbing should take place after the JSON is converted to a string.
Integration with the existing scrubber should be implemented to utilize the secrets list from the associated attempt for scrubbing dataclips with type run_result.
Acceptance requirement:
All sensitive information in the run_result dataclips during the presentation layer should be scrubbed (duh!)
Need to ensure that the page load speed and performance overhead doesn't affect the user or application performance.
Currently, we may expose secrets in the run results in the output dataclip. We will be implementing a rule that limits what the runtime writes to the run_result, but would like an additional failsafe of scrubbing this before displaying it to the user on the presentation layer:
Per the Slack conversation from @taylordowns2000: https://openfn.slack.com/archives/C0394D0GBUN/p1704798081537769?thread_ts=1703149940.085049&cid=C0394D0GBUN
Notes:
Acceptance requirement:
Output and Logs
Output
sectionOutput
(https://demo.openfn.org/projects/4adf2644-ed4e-4f97-a24c-ab35b3cb1efa/attempts/e419d937-0d2c-4df1-b4f0-0e8ffee9140a?r=1f20434d-40b2-4ec8-9a25-f07e54cfda36#output)The text was updated successfully, but these errors were encountered: