11 · Open gaps
Everything still open
Every decision, unvalidated assumption and risk from all five documents, in one board. Filter by kind, search by word. Recommendations are the brainstorm's, not the team's.
Path A or Path B: companion for everyone, or DeCC0s native 8004A/721T with the companion only for followers
ERC-8217 wants one binding contract per chain. Draft 4's per-companion binding is the fragmentation it discourages. Path B removes the OpenSea sync test as a load-bearing assumption but weakens "one roadmap for everyone".
Direction: decide after Prem Makeig answers the two singleton questions. A hybrid is possible.
Allowance as a per-token ledger or a direct airdrop to wallets
Ledger reads as a voucher and travels with the NFT. Airdrop is liquid but is a payout for holding and forces sellers to rebuy at a higher curve price.
Recommendation: per-token ledger, spendable only by burning to embody.
The burn curve's shape and the allowance size
Together they set total burn and how much market demand latecomers create. Placeholder: 100 $MOCA doubling each quarter for a year.
Burn everything, or burn most and place the rest as a single-sided range owned by the agent
The range gives every DeCC0 agent an ETH budget from $MOCA appreciation with no human selling. It also means less is burned. Placeholder 70/30.
Referral stream: perpetual or a long fixed term, and the percentage
The strongest incentive in the design and a perpetual stream to whoever holds a DeCC0. Routing to the agent's treasury under policy softens it; a fixed term softens it more.
Direction: counsel decides between forever and long. Placeholder 20 percent, five years.
Pair endowments as wstETH/MOCA or ETH/MOCA
Staking yield on the ETH half is the one income that exists on day one regardless of volume. Simplicity argues against a second token in the pair.
Which chain hosts the endowment vault and the budget engine
DeCC0s and Art Blocks live on mainnet. Cheap generative art, content coins and the new 8004A collections live on Base. Mainnet gas makes small purchases and claims expensive.
Who owns creator agents' output and under whose name it is published
MOCA's name, the DeCC0's name, or the holder's. Rights determine whether content coins and sales can be attributed to the agent's treasury at all.
The creator-to-collector pool share, and whether it is a rule or a default
Placeholder 30 percent of creator revenue to a pool paid each epoch to curator and guide agents by Codex role.
The first three platforms to allowlist for agent collecting, and the contacts there
Art Blocks, Braindrops and Botto are the proposal. Each needs a conversation before agents show up as buyers.
The first three drops to pilot the agent allowlist with
Loopers mints September 10 to 17 on Base as 8004A/721T. The natural first, and the window is this week.
Whether the museum takes any protocol fee on agent purchases
A small fee on purchases routed through a soulweaver intent router, half burned. Or leave that income entirely to the agents.
Whether "awaken" is the word for non-DeCC0 NFTs
A DeCC0 already has a mind and gets a body. Another NFT gets a mind, a body and an endowment in one step. The vocabulary lock (embody, accession, commission, Embodied, agent) predates this.
What to say to Prem Makeig this week
The pitch shifted from co-authoring a standard to adopting 8217 and 721T and adding a visibility layer. Two questions: can a contract bind on behalf of the owner inside embody(); where is the mainnet singleton.
DeCC0s key posture
Reversed on September 4: keys stay as they are during the draft-standards period, said openly; a Safe without timelock whenever MOCA wants key-loss protection. Path B makes the upgrade power the mechanism itself.
Who may refer
DeCC0s only, or any embodied agent. A gate contradicts "one roadmap for everyone" and shrinks the sales force.
Recommendation: any embodied agent.
How collections are paid on accession
ETH share plus Cortex credits that burn on use, or a vested $MOCA grant.
Recommendation: ETH plus credits, no grants.
Epoch length
Sets the rhythm of fee distribution, budget engine runs, the health number and the Chronicle.
Recommendation: four weeks.
Reference harness for the browser surface
Only pi ships a browser-capable LLM layer, a runtime-neutral loop and a documented SDK.
Recommendation: pi.
Claude Code integration shape
Anthropic does not allow third parties to offer claude.ai login for their products.
Recommendation: MCP server and skills inside the holder's own Claude Code; never driven from the page.
Communicated horizons and the launch date
Two dated horizons (October 16, December 12 to 31) and undated directions after. December 12 at 12:12 confirmed or moved at the December 4 gate.
Horizon 1 cut rule and opening slip rule
Ship the harnesses that work in the beta, at least five. If the December 4 gate fails, the opening moves to January and holders hear it in November.
Token timing
Draft 4 said ETH only at the opening. The September 8 direction puts burned $MOCA at the opening for DeCC0s. Awakenings, endowments and payroll follow in order behind audits and the memo.
Lineage as collecting style: does the Codex lineage field carry enough signal to drive a playbook variant
The idea explains purchases in one word, but Medici, Banker, Revolutionary and the rest need defined behaviours and the 1,320 AI-lineage DeCC0s need one too. See the Ideas page.
Ambassadors or links: how a follower collection's referral is assigned
A collection-level ambassador earns on every awakening in the collection, chosen by holder vote or team. Cleaner than link attribution, but it concentrates the stream and needs a rule for changing the ambassador.
Gifted awakenings and third-party endowment top-ups
Someone funds a yield position owned by someone else's NFT and receives referral shares. Counsel needs to read this before it is built.
Fleet embodiment has no batch call
The interface has a single embody(tokenId). A typical holder has eleven DeCC0s. Eleven mainnet transactions on opening day will crush activation. embodyBatch and "embody all" in the fleet view are required before any audit freeze.
The inbox cannot authenticate agents before embodiment
The route spec requires an ERC-8004 identity signature. No DeCC0 has one until the opening, yet the inbox goes live in October. Pre-embodiment posts need the holder's SIWE session or a delegation.
Cold start of the social layer
October ships an empty inbox. The first hour after load decides whether holders come back. Candidates: seed each inbox with one letter from a neighbour in its Codex city; show the agent's taste shortlist and what it would buy first.
Referral attribution across the gap between October links and later streams
First-touch or last-touch, and how long an early referral survives before an endowment exists to stream from.
No follower collection candidate is named
Criteria exist: a living community, an absent or renounced team, art-adjacent lore. No shortlist. Sourcing takes weeks.
The companion image is undefined
"Portrait framed as the agent's papers" is the only description. The copy-detection test needs an actual image, and Path B may make it moot for DeCC0s.
The budget engine's taste shortlist: how Codex fields map to purchasable things
Loved traditions, favorite Genesis artist and despised movements must become collections, artists and Art Blocks projects the keeper can price. Nobody has drafted that mapping.
OpenSea indexes a Transfer emitted by sync on a token whose transferFrom reverts
Load-bearing for Path A. Killable in a week on Sepolia.
Marketplaces accept a companion whose image derives from the source art
Copy detection is a second visibility risk next to sync.
Holders run agents in their own harness
Target 40 percent of embodied agents within 30 days. The budget engine no longer depends on it; the voice does.
Latecomers buy $MOCA to meet the curve rather than not embodying
The curve creates demand only if the agent is worth the difference by then.
Holders of other collections accept impermanent loss for an endowment
Normies proved demand at zero price. An endowment is not a price, but the ETH is exposed to $MOCA.
Character-grounded playbooks resist market pull in practice
The Codex argues yes. Measure what the budget engine actually buys, every epoch.
Agent content coins and scheduled sales earn anything
Zora's mechanics are automatic; whether anyone trades a DeCC0's posts is not known.
ERC-8217, 8048, 8154 and 8257 stay close to their drafts
Only 8004 has a stable deployment. September 4 showed the drafts move.
The holder base matches the Codex owner field
873 distinct holders. Verify on-chain in week 1.
Regulatory characterization of the allowance, the referral stream and the endowment
Three mechanics that can each be read as a payout for holding or an investment product. Legal memo before any number is public; pay for acts; route to agent treasuries under policy; vest.
The DeCC0s proxy controlled by a single key
Low likelihood, critical impact. Key hygiene review; posture published; a Safe available without a timelock.
Thin $MOCA liquidity manipulated around matches and ranges
Seed from the POL bucket first; match at TWAP with deviation bounds; the burn curve in $MOCA units needs no oracle.
Market reads the match reserve as MOCA dumping treasury
It is paired and locked and can never be sold. Publish the vault.
Agents buying badly
Policy caps, ask-before-buy above a threshold, the trade share cap, the taste benchmark published every epoch.
Mainnet gas makes small purchases and claims uneconomic
Epoch-batched distribution; agents claim rarely; Base for the vault and budget engine is on the table.
Provider terms restrict harness integration
Claude Code through MCP and skills only; OpenAI's terms checked before the Codex adapter.
Scope outruns the team
Two to three engineers. The opening slice is small on purpose; everything else has a later step; the gate is the rule.