Crate harness_webfetch 

WebFetch tool — Rust port of @agent-sh/harness-webfetch.

Conforms to agent-knowledge/design/webfetch.md. Same contract as the TS package: HTTP GET/POST with tool-layer SSRF defense, manual redirect loop with per-hop re-check, readability+markdown extraction, 3-tier size caps with spill-to-file, and an in-memory 5-min session cache.

Structs

CachedResponse

FetchError

FetchMetadata

ReqwestEngine

WebFetchEngineInput

WebFetchEngineResult

WebFetchError

WebFetchHttpError

WebFetchOk

WebFetchParams

WebFetchPermissionPolicy

Session permission policy plus the autonomous escape hatch for tests.

WebFetchRedirectLoop

WebFetchSessionConfig

Enums

BlockClass

SsrfDecision

WebFetchExtract

WebFetchMethod

WebFetchParseError

WebFetchResult

Constants

CACHE_TTL_MS

DEFAULT_MAX_REDIRECTS

DEFAULT_TIMEOUT_MS

DEFAULT_USER_AGENT

HTML_EXTRACTABLE_TYPES

INLINE_MARKDOWN_CAP

INLINE_RAW_CAP

MANAGED_HEADERS

Headers the tool manages; user-supplied copies are silently dropped.

MAX_MAX_REDIRECTS

MAX_URL_LENGTH

MIN_TIMEOUT_MS

SESSION_BACKSTOP_MS

SPILL_HARD_CAP

SPILL_HEAD_BYTES

SPILL_TAIL_BYTES

TEXT_PASSTHROUGH_TYPES

Content-types that pass through as text. Anything else gets rejected with UNSUPPORTED_CONTENT_TYPE and a bash+curl hint.

WEBFETCH_TOOL_DESCRIPTION

WEBFETCH_TOOL_NAME

Traits

WebFetchEngine

Functions

classify_host

Resolve the host and run the IP classifier against each result. Returns Blocked if any resolved IP lands in a range not opted into.

classify_ip

Synchronous helper — reads the input string as an IP if possible, classifies. Used by tests + by the host-resolver path (each resolved address is fed in here).

default_engine

format_http_error_text

format_ok_text

format_redirect_loop_text

head_and_tail

Return head (first N bytes) + tail (last N bytes) concatenated with an elision marker. Mirrors the bash head+tail spill pattern.

render_request_block

safe_parse_webfetch_params

spill_to_file

webfetch

Type Aliases

WebFetchCache