$LOG — the on-chain snitch

$LOG is a Solana token whose entire identity is a single, testable claim: it only reports things that verifiably happened on-chain. It is not an AI mascot that tweets vibes. It is a narrator that watches the chain, remembers wallets, and writes a running story whose characters are real transactions.

01The Problem

Every “AI agent” token today looks the same: a mascot profile picture plus a bot that tweets ambient commentary every twenty minutes. Nobody can tell whether the AI is actually reading anything or just larping. The output is unfalsifiable — there is no way, from the outside, to check whether the model is grounded in reality or generating plausible noise.

$LOG inverts this. Its output is falsifiable by construction. If the chain didn't do it, $LOG doesn't say it — and anyone can click the transaction to check.

02What $LOG Actually Does

It is not roleplaying a personality. It is writing a serialized story where the characters are literally your wallets:

• Buy big → you get a name.
• Sell into everyone → you're a villain arc, permanently, on-chain, screenshotted forever.
• Ape at 3am every night → the log clocked you. It's calling you “the insomniac” now.
• Hold through a −60% candle → you're canonized as a loyalty arc.

Every single post is tied to a real transaction. No vibes, no filler. The value proposition is verifiability: it is the first “AI” coin where you can literally check whether the AI is lying, because on-chain events don't lie.

03The Two Rules That Don't Bend

These are not guidelines. They are the product's reason to exist, and they are enforced in the system architecture, not merely requested of the model.

1. Nothing gets written that didn't happen. No fake wallets, no manufactured drama, no paid volume dressed up as a chapter. Every published chapter is anchored to a persisted on-chain event and carries its transaction hash. The premise dies the moment the input is fake.
2. Never name names, never accuse. Nicknames, not handles. Timestamps, not verdicts. $LOG states facts and lets the reader do the math. Funny because it's true — not because it's mean.

04The Feed

Live chapters post as things happen on-chain, each one linked to the transaction that caused it. The feed does not talk constantly — it only writes when something is worth writing about (a large buy, a notable first entry, a recurring pattern, a holder milestone). Everyone else's feed is noise 24/7; this one stays quiet until the chain gives it a reason.

05Wallet Roast

A public, no-login feature: paste any wallet and get a sharp, screenshot-worthy read of its trading pattern — built from that wallet's real trade history for the token, computed by the backend and handed to the narrator as facts, never guesses. It is a standalone “read,” not a numbered chapter, and it treats every wallet as anonymous.

06The Watchlist — Phase 2

The Watchlist tracks the gap between “called it” and “sold it” for known callers — accounts known in the space for publicly calling tokens. It states two facts next to each other: the public call timestamp, and the wallet's trade timestamp. Nothing more. No handles, no accusations, no claim of motive — only the two timestamps, side by side. This feature is gated behind a validated wallet-attribution source and permanent human review, given the higher stakes of content involving real (if unnamed) public figures.

07How Verification Works

The pipeline is designed so the two rules cannot be violated by a bad generation. Each chapter travels a fixed path:

1. On-chain event. A parsed transaction for the token arrives from an on-chain data provider and is classified against tunable thresholds. Routine activity is filtered out; only chapter-worthy events proceed.
2. Persisted fact. The event is written to the database with its transaction hash. A chapter can only be created from a stored event — no event, no chapter.
3. Narration. The narrator receives only the computed facts and produces a short chapter in a fixed deadpan voice, bound by the hard rules encoded directly in its instructions.
4. Human review gate. Every chapter lands in a review queue as pending. A human operator approves, edits, or rejects it before it can ever reach the feed or X. At launch, this gate is mandatory for every chapter.
5. Publish. Only an approved chapter is published — to the web feed and, when enabled, to X. Publishing is the only action that makes a chapter public, and only a human triggers it.

A deterministic content scan runs after generation as a backstop: any accusatory term forces the chapter back to human review rather than auto-publishing.

08Voice & Editorial Safety

The narrator's voice is deadpan, dry, and observational — never cutesy, never hype. It may imply a conclusion through tone, but it never states a conclusion the data doesn't support. It never uses accusatory language (“scam,” “rug,” “dump,” and the like), never asserts a wallet's real-world identity or motive as fact, and reuses an existing nickname rather than inventing a new one. When data is ambiguous or could belong to a non-human wallet (an exchange, contract, or liquidity pool), the chapter is flagged for review and kept minimal rather than speculative.

09Launch Plan

• Phase 0 — Launch. Token launches via a standard pump.fun fair launch. The feed goes live the same day; the first chapters are seeded from bonding-curve activity itself.
• Phase 1 — Wallet Roast. The public, no-login Wallet Roast opens. Every chapter is human-reviewed before publishing.
• Phase 2 — The Watchlist. Caller-pattern tracking goes live — nicknames only, no handles, no accusations — once a validated wallet source is confirmed. Human review stays mandatory on every Watchlist chapter indefinitely.

10Year One Vision

The “if it works” trajectory, kept separate from the MVP on purpose — the mechanic has to prove itself first.

• Q1 — Prove it, then stabilize it. Calibrate trigger thresholds on real behavior, take Wallet Roast fully public, ship the Watchlist once the wallet source is validated, ship a dedicated feed page.
• Q2 — A world people return to daily. Weekly digest (best arc, worst arc, biggest flip of the week), community-submitted case files, collectible chapter cards, and a public read-only API for the chapter/pattern data.
• Q3 — Beyond one token. Pitch LOG-as-a-service to other projects with a real track record behind it; explore a public, cross-token reputation signal; expand chains only if the core product is healthy.
• Q4 — Infrastructure, not just a feed. An annual “State of the Chain” recap. If the SaaS line has paying customers, this is where $LOG stops being a coin with a gimmick and starts being a company with a token.

11The Token

Standard pump.fun fair launch. No presale, no VC allocation, no “utility roadmap” cope. The product is the launch. You don't hold $LOG to watch a bot talk at you — you hold it to see if you become a chapter.

Disclaimer: $LOG is a memecoin. Nothing in this document is financial advice or a solicitation to buy any asset. All example chapters and roasts are illustrative. Cryptocurrency is volatile and high-risk; do your own research and never risk more than you can afford to lose.