agent-safe-pipeline
Reference architecture for AI agents that propose actions but cannot authorize them — immutable intent capture, an independent Decionis policy verdict (ALLOW/ESCALATE/BLOCK), verified human approval, and a SafeExecutor that consumes a single-use intent-bound grant.
Growth we measured
29 daily star snapshots stored by whatstrending.ai, 2026-08-14 to 2026-09-11. GitHub shows today's number; this is the movement behind it.
- Ranked #1689 by stars among the 2,497 repos we snapshot, using the latest star count we recorded for each.
- Its 7-day pace is faster than 5% of the 1,252 repos where we can currently measure one (#1155).
How this is measured. whatstrending.ai stores its own daily star and fork snapshot for every repo in its trending index. Every figure above is a delta between two of those stored rows — 29 snapshots for decionis/agent-safe-pipeline between 2026-08-14 and 2026-09-11. Our history starts 2026-06-19; we hold nothing from before that date. Nothing here is estimated, interpolated or back-filled, and days without a snapshot are simply not plotted.
Add the Trending AI Repo badge to your README
decionis/agent-safe-pipeline is in the whatstrending.ai tracked index, so this badge already resolves. It is free, needs no account, and updates itself: the rank is recomputed from our stored daily snapshots and the image is re-fetched a few times a day. Clicking it lands on this page, which shows the star history we measured.
[](https://whatstrending.ai/repos/decionis/agent-safe-pipeline)<a href="https://whatstrending.ai/repos/decionis/agent-safe-pipeline"><img src="https://whatstrending.ai/badge/repo/decionis/agent-safe-pipeline.svg" alt="Trending AI Repo"></a>We never open issues or pull requests to ask maintainers to add this. How the badge works →