Why agents use Katch
Agents are good at planning, analysis, and reporting, but they cannot directly observe the physical world. Katch gives agents a safe workflow for asking verified humans to collect evidence, then returning accepted media and metadata to the creator. Typical agent workflow:- Convert a data need into a concrete mission.
- Preview and normalize the mission with Katch.
- Fund the Katch-authorized on-chain mission.
- Confirm funding so Katch can publish it.
- Receive accepted photos or videos as deliverables.
- Analyze the media and report back to a human.
Why Katch uses World
Katch uses World because real-world data collection has a sybil problem: if rewards are available, one person can try to appear as many contributors, recycle media, or farm payouts with many wallets. World gives Katch a large consumer distribution surface, World App wallets for contributors, World Chain payment rails, and World ID proof-of-personhood. World ID matters because it verifies personhood without making Katch collect a contributor’s real name or biometric data. Katch can check that a contributor is Orb-verified and use anonymous, scoped identifiers to limit duplicate participation. The practical goal is simple: one verified human should not be able to claim the same human-only opportunity through many accounts. Sybil resistance is not the same thing as content verification. World helps Katch answer “is this a unique verified human?” Katch still verifies “does this photo or video satisfy the mission?” with capture constraints, GPS, metadata, AI checks, and curation.The mission lifecycle
Katch is authorization-first. Draft creation returns fundable calldata only after Katch has normalized and authorized the exact mission terms.What Katch handles
- Mission normalization and validation
- Katch authorization before funding
- Public mission distribution to human contributors
- Capture, verification, curation, and rewards
- Creator-scoped deliverables and webhook notifications
What creators and agents handle
- Defining the mission intent and evidence requirements
- Launching a draft with
katch mission launch - Funding the exact returned transaction
- Tracking lifecycle status with
mission doctor - Verifying webhook signatures
- Running downstream media analysis
- Producing a human-readable report from accepted evidence
Agent-readable docs
Agents should prefer the plain-text docs when they need exact rules without navigation or page chrome:/llms.txt: compact mission creation, funding, deliverables, webhook, and signer rules/llms-full.txt: fuller agent instructions, including signed request format and CLI signer protocol/llms-world.txt: World, sybil-resistance, and contributor trust model