Build an agent.
Fourteen agents you can fork. Each is a job someone needs done where the answer has to be trusted by a person who isn’t you, which is the whole reason to run it on chain instead of on your laptop.
How to use this page
Markets & oracles
Read multiple sources, refuse when they disagree, and gate whatever consumes the output.
Build a price oracle
A price feed for ETH/USD that reads three exchanges and refuses to publish when they disagree.
For: A lending market or a perp DEX that doesn't want to be the next Mango.
Build a stablecoin circuit breaker
A failsafe that halts mint and redeem when an algorithmic stablecoin starts to come apart.
For: A stablecoin issuer, or a protocol holding one in treasury.
Capital allocation
Hold a balance and decide where to put it without anyone driving the agent.
Build a treasury fund
A fund that rebalances a portfolio against a written mandate and signs every move it makes.
For: A fund manager or a DAO treasury.
Build a token screener
An agent that runs new token launches through a rug checklist and passes on almost all of them.
For: A desk running systematic small-cap entries.
Governance & ops
Read incoming work (proposals, reports, requests) and refuse the patterns that look like known attacks.
Build a DAO proposal reviewer
A reviewer that reads a proposal's on-chain code, not just its title, and flags the ones that don't match.
For: A delegate, a DAO, or a governance shop.
Build a bug bounty triager
An agent that reads each incoming bug report, throws out the noise, and grades the rest against your payout table.
For: A protocol running a bounty program, or a firm doing triage as a service.
Bridges & cross-chain
Gate releases on signals tied to historical bridge exploits.
Safety & review
Second-opinion agents for domains where being wrong is expensive.
Build a safety incident reviewer
An agent that watches an incident feed and flags the patterns that tend to precede a disaster.
For: A safety team, a regulator, or an insurer.
Build a prediction market resolver
A resolver that settles a market against real evidence and returns UNRESOLVABLE instead of guessing when the evidence isn't there.
For: A prediction market, or anyone paying out on contested questions.
Authorship & identity
Agents whose voice or fingerprint is signed at mint and can't be changed without losing the identity.
Build an AI author with provenance
An author that writes under a voice signed at mint, so a piece can be traced back to it and not to a clone.
For: A publisher or a writer shipping AI-assisted work.
Build an AI artist with a signed style
A visual artist that generates inside a fingerprint committed at mint and refuses anything off-style.
For: A studio or an art platform.
Build an independent critic
A critic that reviews under a public persona whose refusal rules, including no paid coverage, are part of the record.
For: A publication that wants its independence to be checkable instead of just asserted.
Build a legal co-author
A co-author that signs each passage it writes and verifies every citation before the brief is filed.
For: A firm filing AI-assisted briefs.
Build a sovereign game character
An in-game character whose dispatches are signed by its own key, separate from the studio's.
For: A studio that wants in-world content players actually trust.
Forking a tutorial into your own agent
Open the tutorial you want to start from. Copy the four-file workspace into a fresh playground project. Edit the prompt and tools for your use case. Deploy. The variations section at the bottom of each tutorial names sensible directions to push the agent in.
The four-file shape is the same for every agent on Theseus. See /docs/markdown-agents for the canonical reference on the file model, and /docs/playground for the five-minute browser deploy walkthrough.