> For the complete documentation index, see [llms.txt](https://docs.hivel.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.hivel.ai/hybrid-hivel-agent/9.-verification-and-summary.md).

# 9. Verification & Summary

### Inspect before you trust

You are not required to take any of the above on faith. The agent supports a download-then-inspect-then-send workflow:

```
# 1. Pull data to local disk only - no Hivel credentials needed, nothing uploaded
./hivel-agent --config config.yml --env-file .env download --output-dir /tmp/snapshot
 
# 2. Inspect exactly what would be sent
gunzip -c /tmp/snapshot/jira/jira_issues.json.gz | jq '.[0]'
 
# 3. Only once satisfied, upload the inspected snapshot
./hivel-agent --config config.yml --env-file .env send --input-dir /tmp/snapshot
```

This lets your security team review the literal payload - the gzip-compressed JSON files that would be uploaded - before any data leaves your network, on the very first run and any time after.

###  Summary - what leaves your network boundary

<table data-header-hidden data-search="false"><thead><tr><th></th><th></th></tr></thead><tbody><tr><td><strong>Category</strong></td><td><strong>Leaves your network?</strong></td></tr><tr><td>Issue/PR/MR/commit metadata (titles, status, dates, authors, labels)</td><td>Yes - this is the product</td></tr><tr><td>Commit/PR/MR message and review comment text</td><td>Yes, unless strip_text_content is enabled</td></tr><tr><td>File paths + added/removed line counts</td><td>Yes</td></tr><tr><td>Source code / file contents</td><td><strong>Never</strong></td></tr><tr><td>Diff or patch content</td><td><strong>Never</strong></td></tr><tr><td>Attachments</td><td><strong>Never</strong></td></tr><tr><td>Your Jira/GitHub/GitLab/Bitbucket/Azure DevOps credentials</td><td><strong>Never</strong> (used locally to authenticate outbound calls only)</td></tr><tr><td>Local git clones used for Bitbucket/Azure DevOps work-metrics</td><td>Never - stays on your infrastructure; only the derived classification numbers are uploaded</td></tr><tr><td>Agent config (config.yml)</td><td>Only if you opt in (send_agent_config), and only with credentials redacted</td></tr><tr><td>Agent logs</td><td>Only if you opt in (stream_logs)</td></tr><tr><td>Raw HTTP traffic incl. credentials (debug.log)</td><td><strong>Never automatically</strong> - written to local disk only; treat as sensitive if you choose to share it manually</td></tr><tr><td>Claude Code usage events (claude_events.jsonl via claudeotel)</td><td>Yes - contents depend on which OTEL_LOG_* flags the customer enabled in Claude Code's own settings; no redact_names_and_urls/strip_text_content equivalent exists for this provider yet.</td></tr></tbody></table>

All data transmission uses HTTPS/TLS. The agent requires no inbound network access at any point.

{% columns %}
{% column %} <a href="/hybrid-hivel-agent/8.-work-metrics-credentials-and-controls.md" class="button primary" data-icon="backward">Work-metrics</a>
{% endcolumn %}

{% column %}

{% endcolumn %}

{% column %}

{% endcolumn %}
{% endcolumns %}