What belongs here
Use this subcategory to report repeatable product behavior that differs from expected functionality. A strong report describes a technical defect rather than a feature request or a usage question, and it gives platform engineers a clear, safe path to reproduce the issue themselves.
What does not belong here
- Credentials of any kind. No API keys, passwords, bearer tokens, session cookies, or user credentials.
- Protected or private material. No student education records, private source files, or unredacted institutional links.
- Suspected security vulnerabilities. Report these directly to security@alkimi.ai rather than posting publicly.
- Account billing and invoice inquiries. Email support@alkimi.ai.
Before you report
- Reproduce it safely. Recreate the issue with a minimal non-sensitive sample document or a fictional prompt scenario.
- Record the specifics. Your precise steps, active Agent settings, expected behavior, actual outcome, and error text.
- Redact before you post. Remove every personal identifier and sensitive institutional detail.
The required report format
Opening a topic here pre-fills your composer with the block below, so in most cases you can just fill it in. It is reproduced here for reference:
## What I expected
[Describe expected behavior]
## What happened
[Describe actual behavior]
## Exact steps to reproduce
1. [Step 1, including model name and reasoning setting]
2. [Step 2, including URL browsing, web search, citations, file uploads, or linked Knowledge collections if used]
3. [Step 3]
## Environment
- Product area: [e.g. Agent Builder, Chat, Studio document processing, Knowledge collection]
- Organization and Workspace: [Redacted]
- Browser and version: [e.g. Chrome 120]
- Operating system: [e.g. Windows 11]
- Source and target formats (for document processing): [e.g. PDF to HTML]
- Approximate time and time zone: [e.g. 14:00 UTC]
## Is it repeatable?
[Every time, intermittently, or only once, plus the smallest non-sensitive reproducing example]
## Redacted error message
```
[Insert redacted error message here]
```
## Safety check
- [ ] No student data or real student scenarios included
- [ ] No credentials, API keys, passwords, tokens, or secrets included
- [ ] No internal URLs or private institutional configuration included
- [ ] No private source documents or restricted policy text included
What happens next
Engineers review formatted reports to isolate defects and verify fixes. Reproduction steps and environment details are what let that work start, so a report carrying both needs less back and forth than one without. Follow the topic if you want to see replies as a report moves.
If an issue occurs only with confidential material, construct a fictional sample with identical formatting to demonstrate the problem safely.
Before you post
Before sharing prompts, configurations, or screenshots, review Community sharing and responsible AI use. Never post student records, personally identifiable information, or credentials.