Connected $GAP acquisition to the embedded wallet.
Added a testnet purchase path that sends $GAP to the Privy wallet while respecting the token's transfer lock.
Log
A chronological record of work worth understanding later, not a mirror of every commit.
Added a testnet purchase path that sends $GAP to the Privy wallet while respecting the token's transfer lock.
Profiles, rankings and shared performance cards can now use names proven by the caller's wallet instead of an unverified label.
Added embedded-wallet entry, durable evidence publishing, call receipts, public challenges and a practice mode isolated from the main weekly record.
Added scheduled reference capture, settlement and final accuracy records so completed rounds can move into durable results without manual stitching.
Active multi-step mint flows now extend while they are in use, stale or superseded actions expire, and validation failures return controlled guidance instead of raw runtime errors.
Continued the control-plane-to-execution integration and tightened transaction lifecycle handling, including first-observation receipt reorg behavior.
Bound the development signer to the execution wallet and fixed the confirmed-wallet preparation transition that had stopped the worker before execution.
Added read-only Ink preflight and controlled target registration so Personal Alpha could validate a real target before execution.
Added durable interactive flows so the owner path could be operated through the control plane instead of command-line-only setup.
The V3 model separates the call window from the Friday-to-Sunday measured gap, allowing calls through the week while preserving the reference boundary used for settlement.
The owner-only worker connected durable jobs to the existing preparation, signing, submission, monitoring and reconciliation boundaries.
The non-production execution test prepared a 64-wallet run around a shared Target Total, with overfill protection preserved.
Established the Robinhood Chain game, weekly rounds, reference-price capture and settlement path.
Hardened the reference-price path so invalid observations and missing required evidence cannot silently produce a result.
Added access-code onboarding, wallet selection and durable mint intents. Telegram creates queued work rather than signing or broadcasting inside the handler.
Configured the static frontend for the verified public deployment without widening its live-purchase authority.
The public assessment path can sync partner proof while live ACP spending remains behind the controlled execution boundary.
Assessment jobs now persist across process restarts, keeping prior runs available to the browser console instead of losing them with runtime memory.
A failed live ACP purchase no longer triggers an automatic paid replacement attempt.
Added source-management flows so approved information can be corrected or deactivated without changing application code.
The bot answers from stored community sources, returns citations and avoids presenting unsupported information as confirmed.
Built the controller, specialist stages and ACP path, then connected the execution console to the real application runtime.
Under the same controlled task conditions, the cold run purchased viability, risk and security for 95. The memory-backed run recalled prior execution experience, stopped after risk and spent 35.
Virtuals ACP job 75660 moved 0.1 USDC into escrow and completed settlement across the provider, evaluator and platform treasury.
A recalled lesson normally allowed security to be skipped. When security became mandatory, the controller kept the useful ordering but overrode the conflicting memory rule for that job.
Built the first Sibyl-backed planning proof, where operational experience persisted across sessions and could alter later specialist procurement.
Hireable service jobs now persist after payment, resume confirmed work after a restart and keep ambiguous settlement failures available for reconciliation instead of losing the job record.
Exposed research, judging and strategy capabilities as independently callable services instead of requiring the full end-user pipeline.
Concept generation moved toward complete competing packages with explicit scoring, rejection reasons and a persisted selection record.
Shortlisted opportunities received their own pain-point research before scoring, with the retained evidence carried into concept generation.
Added GitHub and profile context so opportunity ranking could account for the builder's existing work, strengths and durable preferences.
Connected scouting, builder context, specialist scoring and concept generation into one hackathon-selection pipeline.