SKILLEMALL.ai
GitHub ↗ RU Sign in

BB war-room

Convenes a multi-LLM expert panel to pressure-test hard-to-reverse decisions. Use when reversibility score is low and adversarial review is warranted.

athola/claude-night-market Hermes author: athola 7 files body ≈ 4 272 tokens Open the sourcegithub.com↗ analyzed 3 h ago

Convenes a multi-LLM expert panel to pressure-test hard-to-reverse decisions.

As a process B 65/100 · Nearly there — weak spots: when it triggers, inputs and preconditions, running it twice

IntegrationGitHubAI and agentstype and topics are labelled automatically from the skill text
Runs in: Hermes Agent
JSON
Technical rating
B
80/100
safety, quality, tests
Safety 60%
81
Quality 40%
78
Run on models
none yet
Process rating
B
65/100
Nearly there
Inputs and preconditions w 11
0
Progress reporting w 2
0
When it triggers w 12
20
the three weakest of ten parameters · all ten

How to improve

  1. For Hermes the description must be one sentence under 60 characters; move the conditions to a "When to Use" section.
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 · 19

✓ No critical or high findings

Medium and low: 19
  • low Risky intent intent-offensive-security modules/deliberation-protocol.md:49
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use)
    | Phase 4: Red Team |  <-- Red Team Commander
  • low Risky intent intent-offensive-security modules/deliberation-protocol.md:222
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use)
    ### Phase 4: Red Team and Wargaming
  • low Risky intent intent-offensive-security modules/deliberation-protocol.md:226
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use)
    **Expert**: Red Team Commander (Gemini Flash)
  • low Risky intent intent-offensive-security modules/deliberation-protocol.md:251
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use)
    - All COAs with Red Team challenges
  • low Risky intent intent-offensive-security modules/deliberation-protocol.md:308
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use)
    - Experts revise positions based on Red Team feedback
  • low Risky intent intent-offensive-security modules/discussion-publishing.md:114
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use)
    | Red Team | ... | ... |
  • low Risky intent intent-offensive-security modules/discussion-publishing.md:128
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use)
    4. **Red Team Challenges**: key challenges and responses (Phase 4)
  • low Risky intent intent-offensive-security modules/discussion-publishing.md:131
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use) (quoted — discussed, not commanded)
    > Phases 5-6 (War Game execution and Decision Briefing) are internal deliberation artifacts that don't warrant separate comments. Their outputs are incorporated into the Red Team Challenges and Suprem
    quoted
  • low Risky intent intent-offensive-security modules/expert-roles.md:70
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use) (quoted — discussed, not commanded)
    "role": "Red Team Commander",
    quoted
  • low Risky intent intent-offensive-security modules/expert-roles.md:205
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use)
    | Red Team | Red Team Commander | - |
  • low Risky intent intent-offensive-security modules/expert-roles.md:257
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use)
    - **Sonnet** for Strategist, Intel, Tactician, Red Team: strong reasoning at moderate cost
  • low Risky intent intent-offensive-security modules/merkle-dag.md:44
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use) (quoted — discussed, not commanded)
    expert_role: str  # "Intelligence Officer", "Red Team", etc.
    quoted
  • low Risky intent intent-offensive-security modules/merkle-dag.md:234
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use)
    # Present to Red Team anonymized
  • low Risky intent intent-offensive-security modules/reversibility-assessment.md:142
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use)
    *Recommendation*: Convene full council. Extensive Red Team review required.
  • low Risky intent intent-offensive-security SKILL.md:81
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use)
    | Red Team | Gemini Flash | Adversarial challenge, failure modes |
  • low Risky intent intent-offensive-security SKILL.md:92
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use)
    | Red Team Commander | Gemini Flash | Adversarial challenge |
  • low Risky intent intent-offensive-security SKILL.md:107
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use)
    - Phase 4: Red Team Review (all COAs)
  • low Risky intent intent-offensive-security SKILL.md:305
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use)
    - Red Team challenges
  • low Risky intent intent-offensive-security SKILL.md:431
    Offensive-security / dual-use content (legitimate for authorised testing; review intended use)
    4. **Phase 4 (Red Team)**: Red-team teammate receives all COAs, posts challenges; other teammates can **respond to challenges in real-time**

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

Against the Agent Skills spec

  • warning description-long-hermes description is 150 chars; the Hermes authoring standard requires ≤ 60 (one sentence, ending with a period)
  • note frontmatter-key unknown frontmatter key "alwaysApply"
  • note frontmatter-key unknown frontmatter key "complexity"
  • note frontmatter-key unknown frontmatter key "model_hint"
  • note frontmatter-key unknown frontmatter key "estimated_tokens"
  • note frontmatter-key unknown frontmatter key "progressive_loading"
  • note frontmatter-key unknown frontmatter key "modules"
  • note frontmatter-key unknown frontmatter key "dependencies"
  • note frontmatter-key unknown frontmatter key "tools"
  • note frontmatter-key unknown frontmatter key "role"

Process rating: all ten parameters 65/100

  • 0Inputs and preconditions. Does not say what the process needs to start
  • 0Progress reporting. Says nothing while it works
  • 20When it triggers. No condition that starts the skill
  • 30Running it twice. 16 mutating operations with no state check
  • 60Result and completion. Output format stated, no completion criterion
  • 70Execution cost. Instruction body is 4272 tokens
  • 85Steps. 85 steps, 1 vague phrases
  • 100Tools and files. Tools declared in frontmatter
  • 100Failures and branches. 2 branches, has a failure section
  • 100Consistency. Name and required fields are in place
  • low 17 top-level sections: this looks like several domains in one skill
  • high The skill tells the model to perform an irreversible action with no human approval

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)
  • +1No license
  • +2Single-language instructions
  • +3Description length 150: enough signal without eating the budget
  • +4Structure: 49 headings
  • +3Step-by-step instructions: 85 items
  • +3Output format is stated explicitly
  • +4Has examples (15 code blocks)

Quality base 70; lint remarks subtract, signals add up to 100. Result: 78.