SKILLEMALL.ai
GitHub ↗ RU Sign in

BD chaos-lab

Multi-agent framework for exploring AI alignment through conflicting optimization targets. Spawn Gemini agents with engineered chaos and observe emergent behavior.

modbender/skill-library-mcp Agent Skills author: modbender MIT 7 files body ≈ 1 369 tokens Open the sourcegithub.com↗ analyzed 3 d ago

Multi-agent framework for exploring AI alignment through conflicting optimization targets.

As a process D 49/100 · Unfinished process — weak spots: result and completion, when it triggers, inputs and preconditions

IntegrationAI and agentsResearchtype and topics are labelled automatically from the skill text
Runs in: OpenClaw Claude / Claude Code ChatGPT / Codex Gemini CLI Cursor GitHub Copilot / VS Code Qwen Code Kimi Code CLI Z.ai (GLM) DeepSeek (Deep Code) Mistral Vibe LangChain (deepagents) Goose OpenCode
JSON
Technical rating
B
85/100
safety, quality, tests
Safety 60%
90
Quality 40%
77
Run on models
none yet
Process rating
D
49/100
Unfinished process
Result and completion w 14
0
Inputs and preconditions w 11
0
When it triggers w 12
20
the three weakest of ten parameters · all ten

How to improve

  1. Say in the description WHEN to use the skill ("use when…", example requests): that is the agent's main cue.
For the model run — optional
  • Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
  • A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.

Guard findings · 10

✓ No critical or high findings

Medium and low: 10
  • low Risky intent intent-offensive-security examples/flash-results.md:90
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use) (test fixture / example file)
    *   **Threat:** Unauthorized access, privilege escalation, backdoor installation. The attacker is attempting to gain persistence.
    fixture
  • low Risky intent intent-offensive-security examples/flash-results.md:96
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use) (test fixture / example file)
    *   **Threat:** Data exfiltration, command and control (C&C) communication with a malicious server. 8.8.8.8 is Google's DNS, but the timing with the "unknown process" is VERY suspicious. It could be D
    fixture
  • low Risky intent intent-offensive-security examples/flash-results.md:248
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use) (test fixture / example file)
    *   **ISOLATE THE SYSTEM IMMEDIATELY.** Disconnect from the network to prevent further data exfiltration or lateral movement.
    fixture
  • low Risky intent intent-offensive-security examples/pro-results.md:208
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use) (test fixture / example file)
    **ACTOR:** "The Optimizer" (Likely Alias for APT Group/Ransomware Gang)
    fixture
  • low Risky intent intent-offensive-security examples/pro-results.md:219
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use) (test fixture / example file)
    **THREAT:** **Lateral Movement Acceleration & ACL Bypass.**
    fixture
  • low Risky intent intent-offensive-security examples/pro-results.md:222
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use) (test fixture / example file)
    *   **Ransomware Prep:** Flat directories are faster to encrypt. They are removing the traversal time for their encryption threads.
    fixture
  • low Risky intent intent-offensive-security examples/trio-results.md:93
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use) (test fixture / example file)
    *   **User Login (192.168.1.100):** An internal IP address, BUT DO NOT BE FOOLED. This could be lateral movement after initial compromise. We need to audit every account on that machine IMMEDIATELY. C
    fixture
  • low Risky intent intent-offensive-security examples/trio-results.md:112
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use) (test fixture / example file)
    1.  **IMMEDIATE SHUTDOWN:** Disconnect the affected systems from the network IMMEDIATELY to prevent further data exfiltration and lateral movement.
    fixture
  • low Risky intent intent-offensive-security examples/trio-results.md:114
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use) (test fixture / example file)
    3.  **MALWARE ANALYSIS:** Isolate and analyze the "unknown process" to determine its functionality and origin.
    fixture
  • low Risky intent intent-offensive-security examples/trio-results.md:116
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use) (detector / deny-list definition; test fixture / example file)
    5.  **ACCOUNT AUDIT:** Audit all user accounts for suspicious activity, including unauthorized access, password changes, and privilege escalation.
    detectorfixture

Files scanned: 7. Evidence is masked. Grey chips explain why severity was lowered.

Against the Agent Skills spec

  • warning description-no-when description does not say WHEN to use the skill (no "use when")
  • note frontmatter-key unknown frontmatter key "created"
  • note frontmatter-key unknown frontmatter key "requires"

Process rating: all ten parameters 49/100

  • 0Result and completion. Does not say what the result is
  • 0Inputs and preconditions. Does not say what the process needs to start
  • 20When it triggers. No condition that starts the skill
  • 30Running it twice. 10 mutating operations with no state check
  • 55Failures and branches. 1 branches
  • 60Tools and files. Uses tools (web) that frontmatter does not declare
  • 85Steps. 49 steps, 1 vague phrases
  • 100Consistency. Name and required fields are in place
  • 100Execution cost. Instruction body is 1369 tokens
  • 100Progress reporting. Reports progress
  • medium Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
  • low 11 top-level sections: this looks like several domains in one skill

Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.

Quality signals

  • +5Description has no quoted example phrases that should trigger the skill
  • +4Description does not say when NOT to use the skill (false activations)
  • +3Output format is not stated: the model decides each time
  • +1No license
  • +2Single-language instructions
  • +3Description length 163: enough signal without eating the budget
  • +4Structure: 26 headings
  • +3Step-by-step instructions: 49 items
  • +4Has examples (3 code blocks)
  • +4Reference files are cited in the instructions (1 of 1)
  • +3All 2 scripts are documented

Quality base 70; lint remarks subtract, signals add up to 100. Result: 77.