io.github.ri7in/rule-drift
RuleDrift
Test whether AI agents retain critical instructions as conversations grow.
alive probed 2026-10-15
2tools
327tokens in Claude Opus 5.5
347tokens in Claude Code
0.2 %of a 200k context window
174o200k_base estimate
360 msto initialize
15 msto list tools
- Run it
- npx -y rule-drift@0.1.1 (stdio, probed in a container with the network off)
- Host
- npmjs.com
- Reports itself as
- rule-drift 0.1.1
- Capabilities
- tools
- Instructions
- 163 characters
- Claude count
- Claude Opus 5.5, measured 2026-10-15. Counted again when the tool definitions change. Method.
- Repository
- https://github.com/ri7in/rule-drift
- Registry
- active, updated 2026-08-14 · registry entry
- Reachable
- 1 of the last 1 probe
Most expensive tools
| Tool | Tokens (o200k) | Share of this server |
|---|---|---|
| run_suite | 89 | 51 % |
| validate_suite | 85 | 49 % |
Badge and feed
[](https://mcpcost.ulehla.dev/s/io.github.ri7in/rule-drift)
An entry for every probe that changed the result, the tool count or the token cost: Atom feed. Add it to a feed reader or a Slack RSS app to hear when the server stops answering or its tool set grows.