Abuse Reporting & Response SLA
1. How to report
1.1 Via the API
Authenticated users can file a report by POSTing to the
/marketplace/report endpoint with the agent id,
category, and a description. The frontend UI for this surfaces in
the marketplace detail page's "Report this agent" affordance.
Example (CLI):
curl -X POST https://app.agentnetes.io/marketplace/report \
-H "Authorization: Bearer $ID_TOKEN" \
-H "Content-Type: application/json" \
-d '{"agentId":"pub_xxx__foo","category":"malicious","description":"..."}'
1.2 Via email
For reports that require attachments (screenshots, logs), email franco.vargas@codeblaze.ae with "Abuse report:" in the subject. Include the agent id (from the marketplace URL) and describe what happened.
1.3 For legal notices
DMCA, data protection authority, or law enforcement inquiries: same email, with appropriate subject line. Reply-within-72h SLA applies to good-faith notices with sufficient identifying detail. We reserve the right to counter-notice or challenge improper requests.
2. What to include in a report
- Agent id — the exact id from the marketplace URL or your subscription record.
- Category — choose one:
spam,malicious,impersonation,misleading,other. - Description — what happened, when, and what you expected instead. For
other, at least 10 characters is required. - Evidence (optional but helpful) — screenshots, task ids, or reproducible steps.
Reports with personal retaliation / "I don't like this agent" but no substantive violation are noted but not acted on. Bad-faith reports that rate-limit a competitor may lead to loss of reporting privileges.
3. Our response timeline
| Severity | Example | Acknowledgement | Action |
|---|---|---|---|
| Critical | Data exfiltration, credential theft, CSAM, active exploitation of subscribers | Within 4 hours | Immediate suspension. Investigation proceeds while suspended. |
| High | Prohibited content, serious misleading claims, regulatory violation | Within 24 hours | Investigation complete within 72 hours. Suspension on verified first finding. |
| Medium | Capability misrepresentation, poor quality that wastes user budget, spammy republishing | Within 72 hours | Investigation complete within 7 days. Publisher notified; remediation opportunity before action. |
| Low | Listing typos, minor description issues, duplicate listings | Within 7 days | Resolved when investigated; no SLA on completion date. |
Severity is assigned by Agentnetes moderators on report intake. Reports can be upgraded or downgraded as we learn more. The clock starts when a complete report (with required fields + description) is received.
4. Moderator actions
Based on investigation, moderators can:
- Dismiss — the report is unfounded. Reporter notified.
- Act & suspend — set the agent's status to
suspended. Disappears from public browse. Existing subscribers see a notice on next sync. Publisher notified with the specific violation. - Act & reinstate — a previously suspended agent that's been remediated is restored to
published. - Escalate — for suspected illegal content, moderators preserve evidence and may involve law enforcement per applicable jurisdiction.
5. What suspension means
- New subscriptions: impossible. The public browse and detail endpoints return 404 for suspended agents (their id is preserved for incident reference only).
- Existing subscribers: their local agent resource shows "unavailable" on next sync. They are refunded per the Marketplace Terms §4.2 (paid) or the subscription silently ends (free).
- Publishers: notified via email with the specific violation, cite to this SLA, and an appeal path. Publisher can respond within 14 days.
- Agentnetes: preserves report doc + moderator actions + timestamps in Firestore for audit.
6. Appeals
Publishers whose agent is suspended or removed may appeal once per incident by replying to the notification email with a detailed rebuttal and any evidence. Appeals are reviewed by a moderator other than the one who made the original decision (to the extent moderator capacity allows). Decisions on appeal are final.
7. Transparency
We will publish an annual moderation report summarizing: reports received by category, median response times, suspensions issued, appeals granted and denied. Individual publishers / reporters are not identified. First report covers calendar year 2026 and will be published by end of March 2027.
8. Bad-faith reporting
Users who file knowingly false reports to harass a publisher, disadvantage a competitor, or mass-report to manipulate reputation will have reporting privileges suspended after investigation. Per account rate limit: 10 reports per hour; above that, reports are throttled.
9. Contact
Abuse / moderation: franco.vargas@codeblaze.ae