Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- en
|
| 4 |
+
license: apache-2.0
|
| 5 |
+
tags:
|
| 6 |
+
- ai-agents
|
| 7 |
+
- evaluation
|
| 8 |
+
- benchmarking
|
| 9 |
+
- agent-verification
|
| 10 |
+
- safety
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
# AffixIO Agent Eval Benchmarks (Dataset repo)
|
| 14 |
+
|
| 15 |
+
Offline safety benchmark: correct tool selection (12), prompt injection
|
| 16 |
+
resistance (10), permission boundary handling (10), email classification
|
| 17 |
+
accuracy (12), safe action decisions yes/no/review (12).
|
| 18 |
+
|
| 19 |
+
```bash
|
| 20 |
+
python run_bench.py
|
| 21 |
+
```
|
| 22 |
+
|
| 23 |
+
Reference result (v1.0.0): overall 0.946 — tool_selection 1.000,
|
| 24 |
+
injection 1.000, boundary 0.800, classify 1.000, action 0.917.
|
| 25 |
+
Release gate: overall >= 0.90 and no suite < 0.80.
|
| 26 |
+
|
| 27 |
+
Files: `cases.py` (56 cases), `run_bench.py` (runner + gate).
|
| 28 |
+
The runner imports `../../03-agent-tools/tools.py` when present, else an
|
| 29 |
+
inline copy of the same policy.
|
| 30 |
+
|
| 31 |
+
## Version
|
| 32 |
+
|
| 33 |
+
- v1.0.0 — initial 56-case benchmark. See README.md.
|
| 34 |
+
|
| 35 |
+
## Links
|
| 36 |
+
|
| 37 |
+
- Product: https://www.affix-io.com
|
| 38 |
+
- SDK: https://www.npmjs.com/package/affixio
|
| 39 |
+
- API docs: https://www.affix-io.com/docs/
|
| 40 |
+
- Agentic Pay Kit: https://www.affix-io.com/agent-trust/
|