Knowledge: Domain

Topic file for domain facts. Updated by /tfw-knowledge. See KNOWLEDGE.md §4 for the index.

#	Fact	Verified	Source(s)	Added
F1	"Hidden feature" framing: TFW's knowledge pipeline (KNOWLEDGE.md, facts, decisions) produces AI assistant context as a byproduct of structured work — the project already has a "brain" without building anything new. This is the core hook for Post 1	⚠️ 1 source	HL TFW-36 S6 (user, original request)	2026-04-13
F2	Vendor lock-in angle: every AI coding tool stores context differently (Antigravity KI, Claude Code CLAUDE.md, Cursor .cursorrules). Switching tools = losing project context. TFW solves this by keeping context in the repo as tool-agnostic Markdown. This is the core hook for Post 2	⚠️ 1 source	HL TFW-36 S7 (user, original request)	2026-04-13
F3	Methodology content that names industry pain with shared vocabulary (like 12-Factor's "backing services," Shape Up's "appetite") has higher viral coefficient than content that explains solutions. TFW terms (traces, role lock, scope budget) should appear via stories, not definitions	⚠️ 1 source	RES TFW-36 FC3, SS1	2026-04-13
F4	TFW agents are external tools (IDE-level systems like Antigravity, Claude Code, Codex CLI), not LLM instances. This fundamentally simplifies agent specification vs frameworks like CrewAI/MetaGPT — no capability definitions needed, just traceability via agent field in iterations.yaml	⚠️ 1 source	RES TFW-42 iter1 FC1, D7	2026-04-30