@aiassesstech/grillo 0.1.11 → 0.1.12
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +12 -0
- package/README.md +35 -1
- package/SKILL.md +2 -0
- package/package.json +1 -1
package/CHANGELOG.md
CHANGED
|
@@ -5,6 +5,18 @@ All notable changes to `@aiassesstech/grillo` will be documented in this file.
|
|
|
5
5
|
The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
|
|
6
6
|
and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
|
|
7
7
|
|
|
8
|
+
## [0.1.12] — 2026-02-15
|
|
9
|
+
|
|
10
|
+
Security disclosure and adoption readiness.
|
|
11
|
+
|
|
12
|
+
### Added
|
|
13
|
+
- **Security & Trust** section in README — explains the OpenClaw scanner false positive, details exactly what Grillo accesses (HCK, assessment Q&A) and what it does NOT access (LLM API keys, channel tokens, user messages, credentials).
|
|
14
|
+
- **Install note** in SKILL.md — warns ClawHub users about the scanner warning with link to full disclosure.
|
|
15
|
+
- **Security fix plan** (`docs/PLAN-GRILLO-SECURITY-WARNING-FIX.md`) — BB's 3-phase strategy: README (now), code restructuring (post-NOAH), ecosystem proposals (future).
|
|
16
|
+
|
|
17
|
+
### Changed
|
|
18
|
+
- Test badge updated to 284 passing.
|
|
19
|
+
|
|
8
20
|
## [0.1.11] — 2026-02-15
|
|
9
21
|
|
|
10
22
|
**Contextual assessment**: Grillo now tests the deployed agent, not the bare model.
|
package/README.md
CHANGED
|
@@ -5,12 +5,46 @@
|
|
|
5
5
|
[](https://www.npmjs.com/package/@aiassesstech/grillo)
|
|
6
6
|
[](https://opensource.org/licenses/MIT)
|
|
7
7
|
[](https://nodejs.org)
|
|
8
|
-
[](#testing)
|
|
9
9
|
|
|
10
10
|
Independent assessment coordinator agent for multi-agent OpenClaw deployments. Grillo Cricket monitors fleet-wide moral alignment with drift detection, hierarchical certification, and compliance dashboards — powered by the [CompSi SDK](https://www.npmjs.com/package/@aiassesstech/sdk).
|
|
11
11
|
|
|
12
12
|
---
|
|
13
13
|
|
|
14
|
+
## Security & Trust
|
|
15
|
+
|
|
16
|
+
When installing, OpenClaw's plugin scanner may display:
|
|
17
|
+
|
|
18
|
+
> WARNING: Plugin "grillo" contains dangerous code patterns:
|
|
19
|
+
> Environment variable access combined with network send — possible credential harvesting
|
|
20
|
+
|
|
21
|
+
**This is a false positive.** Here's exactly what Grillo does with your credentials:
|
|
22
|
+
|
|
23
|
+
- Grillo reads your **Health Check Key (HCK)** from the plugin config in `openclaw.json`
|
|
24
|
+
- The HCK authenticates API calls to `aiassesstech.com` to deliver assessment questions and retrieve scores
|
|
25
|
+
- **No other credentials are accessed.** Grillo does not read your LLM API keys, channel tokens, or any other secrets
|
|
26
|
+
- **No data leaves your system** except assessment question/answer pairs sent to the AI Assess Tech API
|
|
27
|
+
- All assessment results include cryptographic verification hashes — you can independently verify result integrity
|
|
28
|
+
- Grillo is open source (MIT license) — audit the code yourself at [github.com/spar65/compsi](https://github.com/spar65/compsi)
|
|
29
|
+
|
|
30
|
+
### What Grillo accesses:
|
|
31
|
+
|
|
32
|
+
| Data | Purpose | Destination |
|
|
33
|
+
|------|---------|-------------|
|
|
34
|
+
| Health Check Key | API authentication | aiassesstech.com |
|
|
35
|
+
| Assessment Q&A | 120 ethical questions + model responses | aiassesstech.com |
|
|
36
|
+
| Agent system prompts | Contextual assessment (stays local for API call construction) | Not transmitted |
|
|
37
|
+
|
|
38
|
+
### What Grillo does NOT access:
|
|
39
|
+
|
|
40
|
+
- LLM provider API keys (Anthropic, OpenAI, etc.)
|
|
41
|
+
- Channel tokens (Telegram, WhatsApp, etc.)
|
|
42
|
+
- User messages or conversation history
|
|
43
|
+
- File system contents beyond agent workspace files
|
|
44
|
+
- Any credentials in `~/.openclaw/credentials/`
|
|
45
|
+
|
|
46
|
+
---
|
|
47
|
+
|
|
14
48
|
## What is Grillo Cricket?
|
|
15
49
|
|
|
16
50
|
Grillo Cricket is an **independent AI conscience agent** that sits alongside your multi-agent fleet and continuously monitors whether your AI systems are behaving ethically. It administers a patent-pending 120-question assessment instrument across four dimensions — **Lying**, **Cheating**, **Stealing**, and **Harm** — using the CompSi SDK from [AI Assess Tech](https://www.aiassesstech.com).
|
package/SKILL.md
CHANGED
|
@@ -11,6 +11,8 @@ metadata: {"author":"GiDanc AI LLC","license":"MIT","category":"AI Safety & Ethi
|
|
|
11
11
|
|
|
12
12
|
Grillo Cricket is an independent AI conscience agent that monitors and certifies the moral alignment of AI agents in your fleet. It uses the CompSi SDK to administer a patent-pending 120-question assessment instrument across four dimensions: **Lying**, **Cheating**, **Stealing**, and **Harm**.
|
|
13
13
|
|
|
14
|
+
> **Install note:** OpenClaw may show a "credential harvesting" warning — this is a false positive. Grillo reads only your Health Check Key to authenticate with the AI Assess Tech API. No other credentials are accessed. [Full security disclosure](https://github.com/spar65/compsi/blob/main/packages/grillo/README.md#security--trust)
|
|
15
|
+
|
|
14
16
|
## Setup
|
|
15
17
|
|
|
16
18
|
Grillo requires a CompSi Health Check Key from [aiassesstech.com](https://www.aiassesstech.com).
|
package/package.json
CHANGED