Platform
Glossary
The terms used throughout these docs, defined once.
- Agent
- A specialist that gathers one kind of evidence — market data, contract powers, public presence.
- Analysis
- One run against one subject, producing a report.
- Artifact
- A downloadable output of a run: a report file, a dataset, a chart.
- Capability
- Something the platform can do, which one or more sources may be able to serve.
- Ceiling score
- A risk score reported as an upper bound because a check could not run. The real score cannot be higher; it may be lower.
- Confirmed
- The check ran and this is the answer.
- Evidence
- One observed fact, with its source and its state. Never AI-generated.
- Execution
- A single run of the pipeline, identified so it can be retrieved later.
- Finding
- An interpreted piece of evidence — what it means, rather than what it is.
- Refuted
- A claim the subject makes about itself does not hold.
- Signal
- A named risk rule and its outcome, with the severity it carries.
- Subject
- The thing being analysed: a token, wallet, contract, website, or the chain itself.
- Summary
- AI-written prose over findings that already exist. Kept strictly separate from evidence.
- Unknown
- The check could not run. Not a negative result, and never displayed as one.
- Watch
- A subject scheduled for re-analysis, producing a change report rather than a repeat.