← ClaudeAtlas

weekly-drift-auditlisted

Run weekly. Detects which context/ files are stale, which foundation-model events happened, which competitor moves matter, and drafts updates for human approval. Implements Living ICP's weekly cadence.
30XCOMPANY/30x-ai-native-gtm-starter · ★ 1 · AI & Automation · score 62
Install: claude install-skill 30XCOMPANY/30x-ai-native-gtm-starter
# Skill · weekly-drift-audit ## What this does Without active maintenance, every `context/` file rots in 30 days. This skill audits the repo weekly, identifies what changed externally that should update the repo internally, and drafts the update for the user to approve. ## How to invoke ``` run weekly drift audit ``` Run every Monday morning ideally; takes 5 minutes of user time. ## Steps Claude takes 1. **Check repo age**: - For each `context/*.md` file, when was it last edited (git log if available, else file mtime)? - Flag any file > 14 days stale. 2. **Scan external events from last 7 days**: **Foundation-model events**: - Any major model release (Anthropic / OpenAI / Google / Meta / Mistral)? - Any pricing change at major providers? - Any new benchmark in relevant vertical? - Any DevDay-class announcement? **Competitor events** (from `context/competitor-radar.md` list): - Funding announcements - Product launches - Pricing changes - Public departures / acquisitions - Notable controversy **Integration / agent ecosystem events**: - New MCP servers in your category? - Cursor / Claude Code / Windsurf integration list changes? - Anthropic Agent Skills directory additions in your space? **Capability obsolescence early-warning** (`competitor-radar.md` Section 2.3 signals): - Did any of the 8 early-warning signals fire this week? 3. **Cross-reference against repo content**: - If a model release happened, doe