A defensible cross-project register of explicit exclusions, phase boundaries, reuse assumptions, and client-provided assets. Every card has a source. Ambiguous “maybe” items were rejected.
Use this as a stop-work shield, not a new contract. “Locked” items are approved/current production decisions. “Current phase” items are excluded now but may reopen later. “Proposal boundary” items are valid estimating assumptions and should be confirmed at award before production relies on them.
48documented relief items
13project workstreams
32locked decisions
16phase / proposal limits
Immediate production ruleFor any green “Locked” card, the default answer is: do not model, texture, rig, animate, or create supporting UI for it unless the project owner supplies a newer superseding decision. Yellow and gray cards remain real scope relief, but verify the current phase or signed scope before using them as a hard stop.
Evidence register
Verified exclusions and limits
Cards are grouped by project. The bold line is the direct production instruction; the quotation beneath it is the source evidence.
No evidence matches this filter.
Research controls
What was included—and rejected
Inclusion standard
Direct “out of scope,” “not supported,” “future phase,” or approved wishlist language.
Client statements that narrow the current pilot.
Explicit reuse or client-provided-asset assumptions that remove net-new 3D work.
Raw Fireflies transcript sentences—not generated meeting summaries—for meeting quotations.
Newer decisions override earlier strawmen.
Rejected or quarantined
TaskUs: an Option A strawman briefly marked items 5–8 out of scope, then Adam replaced it with Option C. It is intentionally absent.
WM: the backing-only inference remains excluded because the RFP did not say so; only the later Adam-directed restrained-pilot asset restrictions are included.
CargoJet: creative-brief exclusions were removed because the accessible trail did not independently establish final human approval.
Waymo, IS4S, AbbVie: discovery questions and AI-generated scope suggestions were not treated as proof.
Pendar spectral graphs and self-test: still pending/conditional in the cited thread, so omitted.
John Deere dirt manipulation and several Halliburton ideas: “probably/most likely/seems” language was not strong enough.