Twitter Mention Blocklist Policy

Public-safe version. Redacted. No handles, no links, no callouts.

Block immediately

• Strongly derogatory replies
• Crypto spam, token shills, wallet/contract junk
• Clear AI slop or reply-bot behavior
• Repetitive low-context spam

Hard triggers

• raw token strings ending in pump
• long crypto-style hashes or contract strings, especially 0x...
• crypto-coded identity markers like BNB, XBT, 0x, Crypto, degen
• low-signal logo or mascot bait from crypto-coded accounts
• community-link pings with no real text
• image-only or link-only drive-bys
• direct hostility with no substance
• repeated emotional blackmail or money-begging

AI / slop signals

One weak signal alone is not enough. Repeated pattern matters.

• generic abstraction soup about agents, workflows, reasoning, interoperability
• fake sage tone with no specifics
• templated agreement or praise across unrelated threads
• same upbeat, shallow reply shape repeated many times in a short window
• canned phrases like:
  • the real unlock
  • the bigger concern is
  • the part nobody mentions is
  • fantastic pattern
  • proper agent infra
  • can provide invaluable insight

Borderline review process

If one tweet is not enough:

• read roughly 8-15 recent profile replies
• inspect follower/following shape
• inspect profile bio
• inspect language fit

Block weight goes up fast when:

• the account is tiny or disposable-looking
• bio pushes coins, tickers, memecoins, or crypto identity
• replies contain cashtags, token names, wallet strings, or contract spam
• the account keeps posting generic praise, fake insight, or thread-detached filler

Prefer mute over block when:

• the account seems human
• the account is tiny
• the content is mostly non-English/non-German
• the replies are specific and technical, not hostile, crypto, or AI-slop

Practical defaults

• If crypto: block.
• If strongly derogatory: block.
• If community-link ping plus crypto chatter in profile replies: block.
• If crypto-coded handle plus silly logo/mascot question: block.
• If image-only or link-only spam: block.
• If repeated emotional blackmail: block.
• If AI-ish: block only when clearly synthetic or repeated.
• If tiny account plus empty hostility: block.
• If tiny account plus human technical chatter in another language: mute.

Redacted examples

• Example A: account drops a ...pump token string in a mention. Block.
• Example B: account asks a silly logo question; profile replies are memecoin chatter. Block.
• Example C: account posts only a community link; profile replies include cashtags and a wallet string. Block.
• Example D: account posts generic AI praise across unrelated threads every minute. Block.
• Example E: account uses direct abuse like fucking drone. Block.
• Example F: tiny account writes a real technical question in another language with specific config details. Mute, not block.