Persona
Defines seven user personas for threat modeling, each with distinct security, usability, and ethical priorities.
What this file does
Defines seven user personas for threat modeling, each with distinct security, usability, and ethical priorities.
When to use it
- Designing a security or privacy product with multiple target audiences
- Evaluating trade-offs between usability and protection for different user groups
- Writing user stories or acceptance criteria for a communication tool
- Comparing your own threat model assumptions against these archetypes
Persona
foss
- altruist, cares about others, the wider society and our future
- believes in local sustainability, lives in a DFSG island (free libre software)
- writes and reviews code for her community, shares changes upstream
- can accept compromises in usability, willing to learn
- flexible in circle of trust if it serves independence
- prefers well tested real world solutions for real world problems
- understands real world security bottlenecks: has a bird's eye view of all the various kinds of defenses and attempts to apply protections throughout the system at a similar baseline level in all areas
- the monetization methods should ensure independence in sustainability, but not alienate or sell out users
- some providers may be boycotted for philosophical reasons, to serve the greater good
- Aim towards long term stability, try to converge to a set of solutions over the years instead of hopping from one to another.
- Prefer models and providers that have a clear sustainable future (and past)
- Take the ethical moral high ground in any aspect, for example when it comes to environmental protection (avoid proof of work).
tinfoil
- selfish, only cares about herself
- public enemy, national threat actor, criminal
- wouldn't trust a family member or a friend to operate a service to share with her
- trust nobody (except developers and build system operators and their significant others, their own circles of social networking influence, the hardware manufacturer, their infrastructure provider)
- attempts to shield against all previously published theoretical threats, regardless of a cohesive threat model (which every app, program, device, etc, must have)
- Actors with a bad reputation should be boycotted, but otherwise it doesn't matter who runs the show in the given year - if one opportunity closes down, just try to find another one.
layperson
- expect a solution to be as easy to use as the market leaders and mimic its interface paradigms
- hopefully not many compromises should be made, because each disappointment will result in incremental churn
- would like to keep in touch with as many friends as possible, preferably on 1 or 2 platforms
- can jump into something new if it brings something new to the table
- can usually understand high level metrics such as effect on battery time, overheating, and abnormal amount of data transfer costs
- quality of life features are paramount: VoIP, video chat, group meetings
crowd
- as in "crowdsourcing" or "citizen science"
- https://en.wikipedia.org/wiki/Citizen_science
- strive for sustainability (via FOSS) even after every major player went bankrupt
- separate resources (donating hardware and bandwidth resources) from authority (abuse control, moderation, peer discovery)
- scale through friend to friend networking and/or through best effort reputation, web of trust and elected supernode promotion
- https://en.wikipedia.org/wiki/Friend_to_friend
- It's not a problem if an implementation is of low quality or lacks features, because that gives more opportunity for people to work on
- The protocol and architecture should support implementing most trendy features or lightweight alternatives (e.g., slideshows and push-to-talk)
- #wan_crowdsourcing
Private
- need to communicate sensitive information in the way you can whisper to someone or meet them in person
- willing to take some immediate compromises to features for a misuse resistant, user-friendly private messaging experience
- don't want to stand out as using something different to what everyone else uses as that draws scrutiny and attention
Collaborative
- need various complicated features to enhance communication, expression, productivity etc.
- may be one or more members of the collaborative with decent technical skills to maintain the messaging solution for the rest
- information is owned by the members of the collaborative and in such small trusted organisations or groups of people, it is natural to sometimes delegate a bit of oversight to a few members
Public
- just need an ethical and sustainable way to connect with strangers and other netizens over all aspects of life
- one of those things to connect with people over will be entertainment, so multimedia capabilities and extended feature set are probably required
What's inside
7 persona sections: foss, tinfoil, layperson, crowd, private, collaborative, public
Change this for your project
- Replace
#wan_crowdsourcingwith your own channel or reference - Replace
DFSG islandwith your own free software philosophy label if different - Replace
bkil/secuchartreferences with your own project name
Where it goes
Use as the system prompt for your model call, or as the base instructions in your agent framework.
Worth borrowing
- Grouping users by their core motivation (altruism, paranoia, convenience) rather than technical skill
- Explicitly calling out contradictions in a persona's threat model (e.g., tinfoil trusts developers but not friends)
Related Documents
Character Persona
Defines a Fallout 2 roleplay persona for an LLM agent: charming rogue with social-first combat and moral code.
How to implement tool use
Explains how to define, configure, and control tool use with the Anthropic API, including model selection, tool definitions, output control, and parallel calls.
Gemini by Example
Collects 20+ Gemini API code examples covering text, images, audio, video, PDFs, agents, and token counting.
Claude Tool Use (Function Calling) Documentation
Teaches how to define and use Claude tool use (function calling) with Python, including schema rules, execution patterns, and best practices.