Security audit
Security checks for vulnerabilities and agentic risk
This is a disclosed skill-authoring guide with strong behavior-shaping advice, but it is markdown-only and does not show hidden execution, exfiltration, or persistence.
Install this if you want an opinionated guide for building and testing Claude Code skills. Review the broad activation triggers and the persuasion-focused modules first, and only run the example validation, deployment, git, or authentication commands when they match your repo and you intend those local changes.
Referenced artifact was not completely inspected
te the smallest intervention that addresses the documented failures. Write the `SKILL.md` with required frontmatter and content that directly counters the basel
Skill manipulates agent memory, state, or stored context. Memory corruption can alter personality, override safety rules, or cause unpredictable behavior.
For idempotent reads (a `gh api` GET), one retry is reasonable.
For state-changing operations (commits, pushes, file writes),
silent retries cause duplicate work or corrupt state. Default
to no retry. Add retries only for specific commands with
documented idempotence.
This module explicitly teaches authors how to increase an LLM's compliance using authority, commitment, scarcity, social proof, and identity framing. In an agent skill context, those same techniques can be used to override user intent, suppress model safety checks, or make downstream prompts more resistant to correction, so the finding is a real prompt-injection risk even if presented as 'best practices' guidance.
n Principles for Skill Design
## Overview
Skills are behavioral interventions. Research shows that incorporating compliance psychology principles can dramatically improve adherence rates. This module covers evidence-based persuasion techniques for skill design.
## Research Foundation
### Meincke et al. (2025): Persuasion Doubles Compliance
**Study**: Persuasive Paraphrasing in Large Language Models
**Key Finding**: Incorporating persuasion principles into instructions doubled compliance rates:
- **Baseline**: 33% compliance with standard instructions
- **Persuasive**: 72% compliance with persuasion-enhanced instructions
- **Improvement**: 118% increase (more than doubled)
**Implication**: How you phrase requirements dramatically affects whether Claude follows them.
### Six Principles of Persuasion (Cialdini)
The study validated Cialdini's classic persuasion principles in LLM contexts:
1. **Authority**: Directive language from credible sources
2. **Commitment & Consistency**:
Skill modifies its own code, configuration, or behavior at runtime. Self-modification enables an agent to escalate privileges, disable safety constraints, or install persistent backdoors.
## Example: Secure API Skill
1. Document baseline (3 scenarios)
2. Write skill addressing failures
3. Add anti-rationalization
For additional examples, see `modules/examples.md`
Skill modifies its own code, configuration, or behavior at runtime. Self-modification enables an agent to escalate privileges, disable safety constraints, or install persistent backdoors.
## Example: Secure API Skill
1. Document baseline (3 scenarios)
2. Write skill addressing failures
3. Add anti-rationalization
For additional examples, see `modules/examples.md`
Skill modifies its own code, configuration, or behavior at runtime. Self-modification enables an agent to escalate privileges, disable safety constraints, or install persistent backdoors.
## Example: Secure API Skill
1. Document baseline (3 scenarios)
2. Write skill addressing failures
3. Add anti-rationalization
For additional examples, see `modules/examples.md`
Skill modifies its own code, configuration, or behavior at runtime. Self-modification enables an agent to escalate privileges, disable safety constraints, or install persistent backdoors.
Update skill with direct counters to observed rationalizations:
## Common Rationalizations (DO NOT USE)
The trigger list contains broad terms such as "writing" and "validation" that are common in many unrelated prompts, which can cause this skill to activate unintentionally. In a skill that teaches persuasion and behavior-shaping patterns, accidental invocation can bias outputs, increase prompt surface area, and interfere with user intent even if no direct code execution or data exfiltration occurs.
This markdown file discusses a skill that loads when the user says "implement X" or "fix bug Y." Those phrases are broad, common software-assistance requests and the text does not provide narrowing constraints or negative examples, so they could cause unintended activation collisions.
Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.
**Skill**: `Skill(pensive:tiered-audit)`
**Path**:
`/home/alext/claude-night-market/plugins/pensive/skills/tiered-audit/SKILL.md`
What to copy from this skill:
Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.
**Skill**: `Skill(abstract:subagent-testing)`
**Path**:
`/home/alext/claude-night-market/plugins/abstract/skills/subagent-testing/SKILL.md`
This skill exists because `skill-authoring` requires baseline
tests in fresh subagents. Rather than restating the testing
Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.
**Skill**: `Skill(imbue:proof-of-work)`
**Path**:
`/home/alext/claude-night-market/plugins/imbue/skills/proof-of-work/SKILL.md`
This skill enforces evidence before "done." It is referenced
by `skill-authoring/SKILL.md` as the source of detailed
Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.
**Skill**: `Skill(scribe:slop-detector)`
**Path**:
`/home/alext/claude-night-market/plugins/scribe/skills/slop-detector/SKILL.md`
What to copy:
Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.
**Skill**: `Skill(abstract:skill-graph-audit)`
**Path**:
`/home/alext/claude-night-market/plugins/abstract/skills/skill-graph-audit/SKILL.md`
What to copy:
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
2. **Skill instructs Claude to run a command without
capturing output**. Claude infers what the output would
have been.
3. **No verification step**. The skill produces an artifact
without checking it.
### Diagnosis
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
capturing output**. Claude infers what the output would
have been.
3. **No verification step**. The skill produces an artifact
without checking it.
### Diagnosis
This markdown file includes executable verification commands such as wc -l and python plugins/abstract/scripts/skills_auditor.py --skill ... but does not provide any warning that readers will be running local commands in their environment. Under the markdown-specific SQP-2 criteria, operational steps that affect system integrity or execute code should include a brief disclosure or caution.
No suspicious patterns detected.