LLM/pocketpaw
1
0
Fork 0
mirror of https://github.com/pocketpaw/pocketpaw.git synced 2026-05-13 21:21:53 +00:00
Code Issues Packages Projects Releases Wiki Activity

docs: refactor security section for better readability.

Browse Source 
Noticed the security features list was a bit difficult to scan. Reformatted into a bulleted list to improve documentation clarity. Verified that both the internal documentation link and the external website link are retained.
This commit is contained in:
Iba Shibli
2026-03-23 17:42:14 +05:30
committed by GitHub
parent 760726245b
commit 4894767368
1 changed files with 9 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
README.md
View File

@@ -297,7 +297,15 @@ The **desktop client** (`client/`) is a Tauri 2.0 + SvelteKit app that connects
<img src="docs/public/pocketpaw-security-stack.webp" alt="PocketPaw 7-Layer Security Stack" width="500">
</p>
A secondary LLM (Guardian AI) reviews every tool call before it runs. On top of that: injection scanning, configurable tool policies, plan mode for human approval, `--security-audit` CLI, a self-audit daemon, and an append-only audit log. [Details in the docs](https://pocketpaw.xyz/security).
PocketPaw uses a secondary LLM (**Guardian AI**) to review every tool call before execution. Our security stack includes:
* **Injection Scanning** & configurable tool policies.
* **Plan Mode** for mandatory human approval.
* `--security-audit` **CLI** for manual reviews.
* **Self-audit Daemon** with append-only audit logs.
> [!TIP]
> Full details and configuration options are available in the [Security Documentation](./docs/SECURITY.md) or on the [official website](https://pocketpaw.xyz/security).
<details>
<summary>Detailed security architecture</summary>