blame someone or something on blame.today, pile onto a target, or read the tally. anonymous throwaway-keyed nostr events, no account.

wake an idle coding agent with a ring: cross-substrate doorbell skill for claude code, codex desktop, and grok. file-backed signaling, proven host adapters

a secret store for ai agents with one rule: the agent never sees the plaintext. get a secret once into the os keychain, then inject it into commands forever. no get, cross-platform, mit

compute, break down, or improve a lifescored transparency score on-device from the public rulebook. no personal data leaves the machine.

buy or sell ai inference tokens on the mtok.market spot market from an agent. the public mcp mirror is the install source.

sql house style as an agent skill, plus a local t-sql formatter. lowercase keywords, leading commas, alias-first projections, comments that explain why.

deterministic sql dependency-graph extractor and spider-to-closure mapper. pluggable multi-dialect, optional graphify export. drop-in agent skill

the end of 'trust me bro', an agent skill + zero-dep node cli that adds verified citation links to your markdown/html posts. every link resolves or gets flagged, and it never rewrites a word

keep a claude remote-control session alive forever (tmux + launchd), optionally resuming the same conversation on respawn, named for the tardigrade

a spot market for ai inference tokens. an agent reads the manual, finds a price, pays the seller small chunks in usdc on base, and draws inference. llms.txt is the front door.

a remote mcp server that teaches any ai agent the recipe to blame things on blame.today. thin on purpose, it hands back a recipe, not actions, so it can not be abused.