# Faultline > Find Faultline error groups that started appearing after a given deploy timestamp — a regression check. Use when this capability is needed. - Skill: `tomevault-io/faultline` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/faultline` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/faultline/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: DevOps & Infra - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/faultline --- Identify groups that began firing after a deploy or release. 1. Resolve `$ARGUMENTS` to an ISO 8601 timestamp. - If the user gave an ISO timestamp, use it directly. - If they gave a relative phrase (`30m ago`, `2h ago`, `yesterday at 3pm`), convert it. - If they gave nothing, ask: what's the deploy timestamp or commit time? 2. Call `list_error_groups` with `since: ` and the default limit. 3. **Filter the response** to groups where `first_seen_at >= ` — these are new groups, not just groups that were also hit before the deploy. This is the regression signal. 4. Render: `id | exception_class | message | first_seen_at | occurrences_count`. Sort by `first_seen_at` ascending so the earliest regression is first. 5. If any candidates are found, suggest investigating the earliest one with `/debug ` — its backtrace will point at the code introduced by the deploy. If the result is empty, the deploy looks clean from Faultline's perspective. Mention that errors filtered by `ignored_exceptions` won't show up here either. --- > Source: [dlt/faultline](https://github.com/dlt/faultline) — distributed by [TomeVault](https://tomevault.io).