Evidence and Token Identity

Public thesis intelligence depends on identity discipline. TradeOS should never treat a symbol alone as enough evidence that it is discussing the correct token.

This matters because many crypto mistakes start with identity confusion. A ticker can be copied, relaunched, bridged, or reused across chains. TradeOS cannot make a token safe by naming it correctly, but identity discipline makes it harder for lookalikes and scam contracts to hide behind a familiar symbol.

Identity key

When available, a token mention should carry:

Field	Why it matters
Symbol	Human-readable ticker, useful but not unique
Chain	Separates assets that share a symbol across networks
Contract address	The strongest practical identifier for an on-chain token
Source timestamp	Shows when the evidence was observed
Source URL or reference	Makes the claim auditable
Confidence	Helps the reader understand how strong the identity match is

For example, a thesis about a Base asset should not be merged with a lookalike token on another chain just because the ticker matches.

Evidence pack standard

The public-intelligence workflow treats the evidence pack as the source of truth. A claim should include:

• what was observed;
• where it came from;
• when it was observed;
• how confident TradeOS is in the observation;
• whether the claim affects thesis support, risk, uncertainty, or a material-change trigger.

The language model can rewrite and organize this information, but it should not add unsupported facts, external rumors, or invented context.

Approved source pattern

Web retrieval is useful, but the safe pattern is constrained retrieval:

1. Fetch from approved sources such as official docs, contract explorers, reputable token lists, DeFi data providers, market data providers, and project announcements.
2. Normalize the result into structured evidence.
3. Let the LLM write only from that evidence.
4. Reject unsupported claims.
5. Keep symbol, chain, and contract together as the identity key.

This keeps the LLM in the role of storyteller instead of source of truth.

Reader-facing language

Public output should avoid vague operational labels. Use labels that tell the reader what kind of research idea they are seeing:

Situation	Reader-facing label
Meme or attention-driven Base assets	Base Meme-Attention Watch
Unknown but interesting tokens	Early Strategic Watch
Weak but potentially interesting candidates	Speculative Watchlist Candidate
High-risk or questionable names	Risk-Watch Thesis

The goal is not to make risky assets sound attractive. The goal is to make the risk understandable.