Safety & Guards¶

MOL has built-in safety primitives — no external libraries needed.

Guard Assertions¶

guard validates conditions at runtime. If the condition is false, execution halts with a clear error.

let confidence be 0.85

guard confidence > 0.5 : "Confidence too low"
show "Passed!"  -- This runs

guard confidence > 0.9 : "Need higher confidence"
-- MOLGuardError: Guard failed: Need higher confidence

Guard Syntax¶

-- With message
guard condition : "error message"

-- Without message (generic error)
guard condition

Guard in Pipelines¶

Guards are powerful inside pipeline definitions:

pipeline safe_answer(question)
  let hits be retrieve(question, "kb", 3)
  let answer be hits |> think(question)
  guard answer.confidence > 0.5 : "Answer quality too low"
  guard len(answer.content) > 10 : "Answer too short"
  return answer
end

Built-in Guard Functions¶

-- Assert minimum value
assert_min(score, 0.7)           -- fails if score < 0.7

-- Assert not null
assert_not_null(data)            -- fails if data is null

Access Control¶

MOL has a built-in security model for AI resource management:

access "mind_core" with "admin"
-- [MOL] 🔓 Access granted: mind_core

Protected Resources¶

Resource	Description
`mind_core`	Core cognitive system
`memory_bank`	Memory storage
`node_graph`	Neural graph
`data_stream`	Data pipelines
`thought_pool`	Thought collection

Access Denied¶

access "secret_resource" with "guest"
-- MOLSecurityError: Access denied to: secret_resource

Type Safety¶

Optional type annotations catch mismatches at declaration:

let x: Number be 42        -- ✓
let y: Text be "hello"     -- ✓
let z: Number be "oops"    -- ✗ MOLTypeError

Error Types¶

Error	When
`MOLRuntimeError`	Division by zero, undefined variable, general errors
`MOLTypeError`	Type annotation mismatch
`MOLGuardError`	Guard assertion failure
`MOLSecurityError`	Unauthorized resource access

Infinite Loop Protection¶

Loops are limited to 1,000,000 iterations:

let i be 0
while true do
  set i to i + 1
  -- Automatically stops after 1M iterations
end

Best Practices¶

Defense in Depth

Use type annotations for critical variables
Use guard to validate pipeline outputs
Use access to protect sensitive resources
Use assert_min / assert_not_null in pipes