Core Principles
Core Principles
What Foundry Does
Operator injects goals; Foundry does
not
define them
Scope
cannot
be expanded by AI
All tasks require explicit acceptance criteria
No task runs without validation
No refactoring without explicit instruction
State persisted after every step
Ambiguity halts execution (with retry mechanism)
Provider CLIs are worker tools, not decision authority
AUTO MODE
is default and mandatory
All outputs are reviewable (diffs, logs)
Anti-Goals (What Foundry Does NOT Do)
Autonomous goal refinement
Speculative task creation
Retry heuristics beyond explicit
retry_policy
AI-only validation (deterministic first)
"Helpful" corrections or fallback behaviors
If in doubt →
HALT
and ask the operator.