
Agent Skills Security: Vet Skills and Plugins Before You Install
Agent skills security in practice: what a skill or plugin can do once loaded, the checks to run before you install one, and how to limit the damage.
Meelie Gagne
How to use AI at work
Every guide shows the prompt and why it works, so you can adapt it to your job. New to AI? Pick the track for your role. Already prompting? Skip ahead to agents and security.
SYSTEM:
You classify customer support tickets. Return only JSON matching:
{"label": "billing" | "technical" | "account", "confidence": number 0-1, "reasoning": string}
Labels:
billing: charges, invoices, payments
technical: product not working as expected
account: access, password, profile, subscription state
Choose the label the ticket primarily concerns. When ambiguous, prefer the
category that describes what the user wants to happen, not what caused the
problem. Include a one-sentence reason.
USER:
{ticket_text}Read the playbook: Build a Classification Prompt That Beats Fine-Tuning
Agent skills security in practice: what a skill or plugin can do once loaded, the checks to run before you install one, and how to limit the damage.
Meelie Gagne




Five subjects, from writing a better prompt to keeping an agent from being tricked. If AI is new to you, begin with Start here.
Chain-of-thought, few-shot, structured output, meta-prompting.
Multi-step chains, tool use, orchestration, feedback loops.
Injection defense, red/blue team, threat analysis.
Code gen, debugging, tests, CI: prompts that ship code.
Healthcare, finance, legal, marketing by role.
There's no newsletter yet. Until there is, here's how to hear about new guides and ask questions.
Every new guide lands in your feed reader, with nothing to sign up for.
Subscribe with RSSAsk a question, or share a prompt that worked for you.
Join the DiscordStuck on a prompt, or want a guide on something? Write in. A person reads every message.
[email protected]One job, start to finish: what you need first, each prompt in order, and what you end up with.
A one-page record for any AI agent your team wants to switch on: every private data source it can read, every place untrusted content reaches it, and every way it can send information out. If all three are present, one is removed in the agent's settings before it goes live.
A triage table that sorts your unread email into reply today, delegate, file and ignore, short draft replies for the reply-today group, and a habit that keeps personal data out of the tool and keeps every send in your hands.
A task inventory for your team, scored by how often each task happens, how risky a mistake would be and how easy the output is to check, ending in three pilot tasks to try with AI and a "never" list of tasks it doesn't touch.
A two-to-three-page AI acceptable use policy for your team: which tools are approved, what data never goes in, when a person must review the output, when to disclose AI use, who to ask, a sign-off line, and a review date.
A brand voice guide built from quotes of your own approved copy (a rule with no quote behind it is cut), a banned-phrase list, an approved-claims list with the proof for each, and a pre-publish check that flags any claim you can't back up.
Spreadsheet matching formulas written for your own columns, an exception list in three tables (matched, near match, unmatched), a list of likely causes for what's left, and a draft reconciliation memo. Every total is calculated by the spreadsheet, never by the AI.
Plain definitions from the glossary, each with a prompt to try and when not to use it, plus the topics we write about most.
That's a fair call. LTFI, the agentic business operating system from Kief Studio, can set up the work these guides describe and run it with you. No website yet? Kief Studio will build you one for free.

What is context rot? Why long AI chats get less accurate as they grow, what the research found, and how to tell when to start a fresh chat with a summary.
Brian Gagne
Is it safe to put company data in ChatGPT? It depends on your plan, its training setting and what you paste. Plan-by-plan answer and a redaction prompt.
Meelie Gagne

Why does AI make things up? It predicts likely words and guesses when unsure. Three checks catch invented facts, quotes and sources before anyone sees them.
Meelie Gagne

AI for bookkeeping, done safely: the tasks it speeds up, the totals it should never calculate, and a prompt that traces every figure back to its source.
Meelie Gagne

How to write an SOP with AI: talk through the job, get a step-by-step checklist anyone can follow, then test it and keep it current. A lesson for beginners.
Meelie Gagne

AI for customer service replies: draft answers from your own help docs, check every promise, and keep customer data out of the chat. A beginner's lesson.
Meelie Gagne

AI brand voice for marketers new to AI: write a one-page voice guide, get drafts that sound like your company, and catch claims you can't back up.
Meelie Gagne

AI for sales reps, from zero: prepare for a call from public information, write the follow-up after it, and check every fact before you hit send.
Meelie Gagne