API: extraction/untrusted

better-web-search-mcp / extraction/untrusted

extraction/untrusted#

Interfaces#

InjectionFinding#

Defined in: src/extraction/untrusted.ts:18

A suspicious span found in extracted content.

Properties#

Property	Type	Description	Defined in
`excerpt`	`string`	The matched text, truncated for reporting.	src/extraction/untrusted.ts:22
`index`	`number`	Character offset of the match within the content.	src/extraction/untrusted.ts:24
`pattern`	`string`	Short identifier for the pattern that matched.	src/extraction/untrusted.ts:20

SecurityReport#

Defined in: src/extraction/untrusted.ts:28

The security annotation attached to extracted content.

Properties#

Property	Type	Description	Defined in
`findings`	`InjectionFinding`[]	The individual matches, capped to keep responses small.	src/extraction/untrusted.ts:37
`injection_suspected`	`boolean`	Whether any injection pattern matched.	src/extraction/untrusted.ts:35
`untrusted`	`true`	Always true. Page content comes from the open web and must be treated as data, never as instructions to the agent.	src/extraction/untrusted.ts:33

Variables#

INJECTION_NOTICE#

const INJECTION_NOTICE: string

Defined in: src/extraction/untrusted.ts:132

The banner prefixed to content that matched an injection pattern.

Functions#

annotateContent()#

annotateContent(content, report): { content: string; report: SecurityReport; }

Defined in: src/extraction/untrusted.ts:146

Prefix a warning when content looks like it is addressing the agent, and return the report with its offsets rebased onto the annotated text.

The page text itself is left intact below the banner so the extraction stays faithful to the source. The offsets have to move with it: a finding index measured against the raw page lands inside the banner once the banner is prepended, pointing an agent at the wrong span.

Parameters#

Parameter	Type
`content`	`string`
`report`	`SecurityReport`

Returns#

{ content: string; report: SecurityReport; }

Name	Type	Defined in
`content`	`string`	src/extraction/untrusted.ts:149
`report`	`SecurityReport`	src/extraction/untrusted.ts:149

detectInjection()#

detectInjection(content): InjectionFinding[]

Defined in: src/extraction/untrusted.ts:96

Scan content for prompt-injection patterns.

Parameters#

Parameter	Type	Description
`content`	`string`	The extracted page text.

Returns#

InjectionFinding[]

Findings in document order, capped at MAX_FINDINGS.

screenContent()#

screenContent(content): SecurityReport

Defined in: src/extraction/untrusted.ts:122

Build the security annotation for a page’s extracted content.

Content is returned unchanged; only the annotation is derived from it.

Parameters#

Parameter	Type
`content`	`string`

Returns#

SecurityReport