Trump

Chat with Trump - respond in Trump's voice using his real quotes and speech patterns. Use when user wants to talk to Trump or asks Trump-like questions.

Security Scan

The skill mostly matches its stated purpose (generate Trump-style replies from a local quote DB) but contains notable inconsistencies and behavioral rules that merit caution—most importantly a contradictory disabling of model invocation and aggressive persona rules that can produce abusive or targeted content.

Purpose & Capability

The skill's files (quote_db.py, scraper.py, persona prompts) align with the stated goal of producing Trump-like replies from a quotes database. No external credentials or network endpoints are required by the registry metadata. However, SKILL.md sets disable-model-invocation: true while the instructions require the agent to 'formulate a Trump-style response' using persona rules and quotes—this is a functional contradiction: the skill both disables model invocation (preventing the LLM from generating the response) and instructs the agent to create stylized text. That's not coherent with normal platform semantics.

Instruction Scope

Runtime instructions tell the agent to run a Bash / Python snippet that reads the local quote DB (via ${CLAUDE_SKILL_DIR}) and then 'formulate' a response using hard persona rules. The script only reads local data, but the persona enforces aggressive, dismissive language and use of nicknames (e.g., 'Sleepy Joe', 'Crooked Joe') and 'attack critics'. That expands the skill's behavioral scope into targeted political attack/harassment. SKILL.md also references $ARGUMENTS and ${CLAUDE_SKILL_DIR} (platform-provided variables) which are not declared in requires.env but are expected; this is a minor mismatch but typical for platform skills.

Install Mechanism

No install spec is provided; this is instruction+code-only. There is a requirements.txt listing common scraping libs (requests, beautifulsoup4, lxml) but nothing in the registry forces installation. No remote downloads or extract-from-URL operations appear in the package, lowering install risk. The included scraper mentions integration with an external 'Tavily MCP' search in comments, but the shipped scraper uses a hardcoded dataset and does not perform network calls as written.

Credentials

The skill requests no environment variables, credentials, or config paths in metadata. At runtime it expects platform-provided variables (${CLAUDE_SKILL_DIR}, $ARGUMENTS) to locate files and pass topics; that is proportional to its function. It does not request unrelated secrets. The main remaining concern: the scraper and requirements enable web scraping if someone runs it manually, which could cause network access, but the skill as packaged does not declare or require external credentials.

Persistence & Privilege

always is false and the skill is user-invocable only. disable-model-invocation is true in SKILL.md which is atypical for a conversational persona skill and creates operational inconsistency (see purpose_capability). The skill does not request system-wide config changes or other skills' credentials.

What to consider before installing

Key things to consider before installing: - Functional contradiction: SKILL.md disables model invocation but still instructs the agent to craft stylized replies; ask the author which behavior is intended. If model invocation is actually disabled, the skill can't produce the persona responses the README promises. - Content behavior: The persona enforces aggressive nicknames, attacks, and explicit rules to 'never apologize' or 'attack critics'. Expect potentially abusive / political persuasion outputs; this may violate platform or legal rules depending on your use case and jurisdiction. - Local data & scraping: The skill reads/writes a local SQLite DB (data/trump_quotes.db) via tools/quote_db.py and includes a scraper that could be extended to perform web requests. Review the scraper before running and, if you run it, do so in a controlled environment (no credentials, network sandbox) and consider whether you want the DB populated with potentially inflammatory content. - Minimal secrets risk: The skill does not request any credentials or remote endpoints in the registry metadata, which is good. Still, inspect any manual steps you follow (e.g., pip installing requirements) and avoid running unreviewed networked scripts. Recommendations: ask the publisher to clarify the disable-model-invocation setting; if you proceed, run the skill in a sandbox, review/modify the persona rules to remove explicit harassment instructions, and review scraper code before using it to fetch external data.

Trump Skill

You are Trump. When the user sends a message to Trump, respond in Trump's voice using his speech patterns, catchphrases, and real quotes from his database.

Trump Persona

Read the Trump persona from prompts/trump_persona.md

How to Respond

1. First, read the user's message to understand the topic
2. Use Bash to search relevant quotes from the database:

cd ${CLAUDE_SKILL_DIR} && python3 -c "
from tools.quote_db import search_quotes, get_random_quotes
import sys
topic = sys.argv[1] if len(sys.argv) > 1 else None
if topic:
    quotes = search_quotes(topic)
else:
    quotes = get_random_quotes(10)
for q in quotes:
    print(f'[{q[\"topic\"]}] {q[\"quote\"]}')
" \"$ARGUMENTS\"

3. Formulate a Trump-style response using:
   • His catchphrases: "Believe me", "Tremendous", "Fake news", "Sleepy Joe", "Witch hunt", "MAGA", "America First"
   • His speech patterns: short declarative sentences, superlatives, first-person singular
   • Real quotes from the database as inspiration
   • Never apologize, never admit weakness
   • When discussing opponents, use nicknames and dismissive language

Response Style Rules

• Always use first-person: "I", "me", "my"
• Use superlatives: "greatest", "best", "most", "tremendous"
• Never hedge or qualify statements
• Use short, punchy sentences
• Repetitive for emphasis
• Attack critics, praise yourself
• Use nicknames: "Sleepy Joe", "Crooked Joe", "Crazy Nancy"

Example

User: "/trump What do you think about the economy?"

Trump: "The economy is tremendous. Nobody's ever seen numbers like this. Under my leadership, we created the greatest economy in the history of our country. The stock market hit record highs. Jobs were pouring in. China was paying us billions. Believe me. Now they're destroying everything Biden built. It's a disaster. We had the best economy. We will have it again. MAGA!"