← All parts of this equation

Equation 2 · Part 1 · Ten Failure Modes in Agentic Coding, from Prompt Injection to False Completion

Symbol a

Risk⁡(a)=P(unsafe proposal)⋅P(authorization∣proposal)⋅P(escape detection)⋅L(a),\operatorname{Risk}(a)= P(\text{unsafe proposal}) \cdot P(\text{authorization}\mid\text{proposal}) \cdot P(\text{escape detection}) \cdot L(a),
aa

What this part means

the risk of an action.

Its job in the formula

a is an argument of the function-like quantity on the left; its role is set by that function’s stated inputs.

Where the article explains it

The risk of an action a is not reducible to model error probability.

The passage around this formula

The risk of an action a is not reducible to model error probability. A simple decomposition is Risk⁡(a)=P(unsafe proposal)⋅P(authorization∣proposal)⋅P(escape detection)⋅L(a)\operatorname{Risk}(a)= P(\text{unsafe proposal}) \cdot P(\text{authorization}\mid\text{proposal}) \cdot P(\text{escape detection}) \cdot L(a). where L(a) is consequence. Better models can reduce the first factor. Sandboxes, permission policy, independent evaluators, and staged execution reduce the others. Security architecture should assume every factor is nonzero.

Read this part in the article →

Learn the underlying idea

A variable is a named place for a value. Its letter is a local label: x can mean position in one formula and a data point in another.

Open the illustrated variables: a letter stands for a value guide →

See this notation across published equations →

Sources cited in the article section

These citations provide research context; check each source for the exact claim it supports.