Security incident log and post-incident summary drafts
During and after a security incident, an agent keeps a timeline from chat messages and tickets and drafts a factual post-incident summary for the incident lead to correct, complete, and approve.
The use case
When something goes wrong, such as a compromised email account or a lost laptop, small teams focus on fixing the problem and rarely write down what happened. Later, insurers, customers, or auditors ask for a timeline, and nobody remembers the details. The agent watches a dedicated incident channel and the related ticket, records each message, action, and decision with timestamps, and keeps a running timeline. After the incident is closed, it drafts a summary covering what happened, what was affected, actions taken, and open follow-ups, using only what was recorded. The incident lead corrects the draft, adds missing facts, and decides who receives it. Any notification to customers, regulators, or insurers is decided and written by the business, with legal advice where needed. The agent does not contact anyone outside the team.
Example workflow
- When an incident is declared, a dedicated Slack channel and a ticket in Jira or Linear are created.
- The agent reads messages in the channel and updates on the ticket and records events in a timeline in Google Docs.
- It notes decisions, actions taken, and open questions, each with a timestamp and the person involved.
- After the incident is closed, it drafts a summary of what happened, affected systems and data, actions, and follow-up items.
- The incident lead reviews the draft, corrects facts, adds context, and approves the final version.
- Follow-up items are copied into the task tracker with owners and due dates.
Tools
- Slack
- Jira or Linear
- Google Docs
- Claude
- Notion
Risks and guardrails
- The draft speculates about causes or impact — guardrail: the agent writes only what was recorded and marks gaps as unknown.
- Sensitive details, such as passwords or personal data, are copied into the timeline — guardrail: the agent redacts credentials and personal data, and the timeline is stored in a restricted folder.
- The summary is sent to customers or regulators without review — guardrail: external communication is decided by the business, with legal advice where needed, and the agent never sends it.
- Messages outside the channel are missed — guardrail: the incident lead adds key facts from calls or other channels to the timeline.
Related recipe
Support Ticket and Review Triage with Policy-Cited Draft Replies: An agent tags every helpdesk ticket and public review by intent, urgency, and sentiment, routes it, and drafts a reply that cites your help center. People approve refunds and all public replies.
Want it set up for you? Request a quote.
Sources
Written by Agents Lookup. Credits and official pages for the tools this use case names:
- Slack MCP server (official page): docs.slack.dev/ai/slack-mcp-server (checked 2026-11-09)
- Linear MCP server (official page): linear.app/docs/mcp (checked 2026-11-09)
- Claude (official page): claude.ai (checked 2026-11-09)
- Notion MCP server (official page): developers.notion.com/guides/mcp/overview (checked 2026-11-09)