You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Commit 6ca98f3
Browse filesBrowse the repository at this point in the historyBrowse files
Copy file name to clipboardExpand all lines: .github/actions/unplug-scan/README.md
+6-2Lines changed: 6 additions & 2 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -2,6 +2,10 @@
2
2
3
3
Scan **agent-related files changed in a PR** with the unplug-ai regex Guard. Intended for repos that ship `AGENTS.md`, `.cursor/rules`, or MCP client configs.
4
4
5
+
**External repos:** prefer the published Marketplace action **[UnplugAI/unplug-scan-action@v1](https://github.com/UnplugAI/unplug-scan-action)** (PyPI install, semver tags).
6
+
7
+
This composite action lives in the monorepo for local SDK development.
8
+
5
9
## Usage in this repo
6
10
7
11
Already wired in [`.github/workflows/pr-scan.yml`](../workflows/pr-scan.yml).
@@ -23,7 +27,7 @@ jobs:
23
27
with:
24
28
base-ref: main
25
29
install-mode: pypi
26
-
unplug-version: ">=0.3.1"
30
+
unplug-version: ">=0.4.0,<0.5"
27
31
```
28
32
29
33
## Inputs
@@ -34,7 +38,7 @@ jobs:
34
38
| `python-version` | `3.12` | Python version |
35
39
| `working-directory` | `sdk` | Path to SDK tree when `install-mode: local` |
36
40
| `install-mode` | `local` | `local` (uv sync) or `pypi` (install from PyPI) |
37
-
| `unplug-version` | `>=0.3.1` | Version constraint for PyPI install |
41
+
| `unplug-version` | `>=0.4.0,<0.5` | Version constraint for PyPI install |
Copy file name to clipboardExpand all lines: README.md
+1-1Lines changed: 1 addition & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -70,7 +70,7 @@ Try it without installing anything: [live demo](https://huggingface.co/spaces/Un
70
70
| ML span model `Guard.with_tiny()`|**Preview** ([unplug-tiny-v1](https://huggingface.co/Unplug-AI/unplug-tiny-v1)) |
71
71
| Sliding-window long documents + streaming scan |**Included**|
72
72
73
-
Regex-only doc-level detection reaches roughly **F1 0.36 / recall 0.23**on held-out attacks: fine as a first line, not sufficient alone. The ML span model's measured per-axis numbers (including failures) are on the [model card](https://huggingface.co/Unplug-AI/unplug-tiny-v1).
73
+
On the neuralchemy prompt-injection set, regex-only detection reaches **F1 0.56 / recall 0.39**— a fast first line, not sufficient alone. Adding the ML span model (`Guard.with_tiny()`) takes that to **F1 0.99 / recall 0.98**, and lifts recall on *indirect* injection from **0.05 → 0.91**. False-positive rate stays under 1% on the injection set (2.1% on a separate hard-benign corpus). Full tables, methodology, and honest caveats: [`sdk/docs/BENCHMARKS.md`](sdk/docs/BENCHMARKS.md). Per-axis model metrics (including failure modes) are on the [model card](https://huggingface.co/Unplug-AI/unplug-tiny-v1).
0 commit comments