Draft a Process Intelligence issue
Draft first, do not submit by default. There is no fixed destination repository. Use only the user's supplied information; this workflow performs no diagnostic collection.
- Capture the symptom, reproduction, expected/actual behavior and sanitized error code. Include versions only when supplied. Ask one focused question at a time for missing essentials; do not ask again for supplied facts or invent missing details.
- Do not collect or read logs, profiles, credentials, transcripts or unrelated environment data. Do not run authentication, credential checks, shell commands or business calls to prepare a report. Remove tokens, claims, usernames, tenant/environment/user identifiers, raw errors, tool inputs/results, business values, machine paths and internal URLs from the draft.
- Show the full preview: intended destination and visibility, title, symptom, reproduction, expected/actual behavior, supplied versions and safe error codes. Include supplied session/request correlation IDs only after the user reviews them; a private destination is not permission to include sensitive data. Allow corrections. Missing evidence does not prove success, failure, cancellation or a hung backend.
- Stop at the draft. Any later submission requires explicit approval of the exact destination and final content, through the host's supported issue-creation tool. Do not automatically submit, put the report in a URL query, upload files or silently check credentials.