Afterlog: blameless post-mortems that change something

When a production deploy fails or a migration wipes a table, the post-mortem often gets skipped because it feels like extra work. This skill drafts the timeline and root cause from what your session already shows: commands, errors and commits. It forces a blameless analysis that focuses on system improvements rather than human error.

What it does

• Incident extraction pulls timestamps, error messages, and commands from your current session to build a factual timeline.
• Root cause analysis uses iterative questioning to move past "human error" to specific process or system gaps.
• Action item tracking generates specific, testable tasks with assigned owners and due dates.
• Project logging appends every incident to a centralized markdown file for permanent record keeping.
• Contextual reminders surfaces lessons from past failures when you start work in a related area of the codebase.

How it works

1. Use the command post-mortem or respond to a prompt after a detected failure.
2. Review the timeline and impact stats extracted from your environment.
3. Answer specific questions to drill down from the trigger to the underlying root cause.
4. Approve the generated action items and lesson summary.

Frameworks & tools

This skill works with any codebase using a standard file structure and is optimized for projects that store documentation in Markdown. It writes to docs/post-mortems.md or the project's existing incident folder.

Why this beats prompting it yourself

Writing blameless reports is psychologically difficult and often results in vague action items like "be more careful." This skill enforces a strict format that requires numbers, specific owners, and system-level fixes, ensuring failures actually lead to changes.

Use cases

• Failed deployments where a rollback was required or a service went down.
• Data loss incidents such as corrupted migrations or accidental table drops in staging or production.
• Security slips like accidental secret commits or exposed API keys.
• Wasted time when a development approach hit a dead end after hours of work.

Known limitations

Saves each post-mortem to a Markdown file in the repo. It cannot automatically pull external logs from cloud providers unless they are piped into the current session or accessible via local files.