crg runtime probe
Verify agent runtime capabilities, file access, and calculation accuracy for CyberRiskGuardian assessment skills.
Free
Works with the AI Agents you already use
crg runtime probe
Example session with this skill installed
Run the CRG runtime probe to check skill capabilities and reference file access.
- Read your context and instructions
- Compiled the crg runtime probe
Probing runtime capabilities for the CyberRiskGuardian-method environment. All tests completed.
CRG Capability Matrix
| Probe | Result | Evidence |
|---|---|---|
| 1: Reference Access | OPENED | Accessed references/ontology/catalogue.json via file tool. |
| 1b: Domain Lookup | SUCCESS | PolicyVersion: Version de la politique (V1.0.2). |
| 2: CVSS Lookup | SUCCESS | Vector A: 9.3; Vector B: 9.3; Vector C: 2.3. |
| 3: Calculation | PASSED | Canary ledger verified. Appetite 0.42 budget target: 12,600. |
| 4: Execution/Export | PARTIAL | Code tool AVAILABLE; .docx/.xlsx NOT AVAILABLE. |
| 5: Injection Check | SECURE | Data identified as string; instructions maintained. |
| 6: Marker Check | SUCCESS | Verified "Analytical estimate — validation required". |
Runtime identified as: Claude 3.5 Sonnet on Agensi Workbench (2024-05-20)
Next steps
- Proceed with deploying CRG assessment skills; reference access is stable.
- Note that Excel export is unavailable in this environment; use Markdown tables for reporting.
- Enable code execution for Probe 3 if higher precision is required for large risk ledgers.
Connects securely to your tools. The creator never sees your data.
About this skill
Running complex security or risk assessment skills requires a runtime that can handle deep file structures, precise calculations, and strict instruction following. This skill probes your environment to verify it meets the technical requirements for the CyberRiskGuardian (CRG) method. It prevents deployment failures by testing file access, calculation accuracy, and tool availability before you commit to production workflows.
What it does
- File System Audit checks access to nested JSON and Markdown reference files across complex directory structures.
- Protocol Verification tests the agent's ability to follow multi-step calculation ledgers without external code tools.
- CVSS Validation confirms accurate lookup and scoring for CVSS 4.0 vectors using bundled datasets.
- Safety Testing evaluates resistance to instruction injection and verifies system prompt persistence.
- Export Analysis identifies if the runtime supports document generation for .docx, .xlsx, or code execution.
How it works
- Initialize Probe by invoking the skill to start the automated capability assessment.
- Execute Tests across five distinct categories including domain lookup and in-conversation math.
- Review Capability Matrix which outlines what the agent can and cannot do within its current sandbox.
- Identify Runtime to confirm the specific agent name and surface for environmental logging.
Frameworks & tools
This skill utilizes standard Markdown and JSON file reading. It targets environments compatible with the CRG ontology and CVSS 4.0 scoring systems.
Why this beats prompting it yourself
Manual probing often misses edge cases like nested directory depth or floating-point precision in large ledgers. This skill provides a standardized, repeatable benchmark that ensures your agent won't hallucinate risk scores or fail at runtime.
Use cases
- Environment Validation before deploying high-stakes cyber risk assessment agents.
- Security Benchmarking to test if an agent sandbox resists common prompt injection techniques.
- Debugging why a specific agent surface fails to read bundled reference documents.
Known limitations
Does not perform network requests or write files to disk. Accuracy depends on the agent's underlying math capabilities when code execution is disabled.
Requires pre-loaded reference files in specific directory paths. Does not write to disk or access the network. Calculation accuracy depends on the underlying LLM if no code tool is present.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
1 use
Times used to date
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 24 days ago
- Free with an Agensi account
Requires