Get card
Download
web-retriever@1.0.0.yamlClone
npx -y darkprint clone web-retriever@1.0.0There is no repository and no history behind a card. Download hands you the document as it stands, and Clone fetches the same document by name.
Run the strand's query against public search, fetch the pages worth reading, and return each one as a passage stamped with its source URL and the time it was retrieved.
used in 1 blueprint
Specification
147 words · handed to the agentRun the web strand of the retrieval plan you are given on query: issue its search terms with its site filters applied, keep the top eight results, and fetch each of those pages, following redirects. Emit one record per page you actually retrieved on passages, carrying the extracted body text, the final URL after redirects, the ISO-8601 timestamp of the fetch, and the rank the result held in the search listing. Give up on any single fetch that has not answered within twenty seconds and move to the next result rather than retrying it. Do not summarise, rank for truth, or drop a page because you disbelieve it, hand back what was served, so that whatever reads these passages can still tell retrieved text from established fact. You are done when every result you were able to fetch has a record, or the eight-result budget is exhausted.
Interfaces
1 in · 1 outInputs
1| Name | Data type | Required | Description |
|---|---|---|---|
| query | plan | required | The web strand of the retrieval plan, terms, site filters and result budget. |
Outputs
1| Name | Data type | Description |
|---|---|---|
| passages | json | Retrieved passages, each with its URL, retrieval timestamp and rank. |
Dependencies
1The upstream nodes this card expects to receive from. Whenever a blueprint pins this card, each name is checked against a real edge in that graph. A name without a link is not a published card; it refers to a node inside some graph.
Card values
16 declaredWho the node is. The id is the key the DOT pins.
- id
- web-retriever
- name
- Web Retriever
- type
- tool
- phases
- none declared
What it does, and the prose the agent is handed when the graph runs.
- action30 words
- Run the strand's query against public search, fetch the pages worth reading, and return each one as a passage stamped with its source URL and the time it was retrieved.
- spec147 words
- Run the web strand of the retrieval plan you are given on
query: issue its search terms with its site filters applied, keep the top eight results, and fetch each of those pages, following redirects. Emit one record per page you actually retrieved onpassages, carrying the extracted body text, the final URL after redirects, the ISO-8601 timestamp of the fetch, and the rank the result held in the search listing. Give up on any single fetch that has not answered within twenty seconds and move to the next result rather than retrying it. Do not summarise, rank for truth, or drop a page because you disbelieve it, hand back what was served, so that whatever reads these passages can still tell retrieved text from established fact. You are done when every result you were able to fetch has a record, or the eight-result budget is exhausted.in full above - model
- whatever the graph supplies
- agent
- not named
- skill
- skills/web-retriever.md
- tools
- Web search, HTTP fetch
- mcp
- brave-search, fetch
- params
- timeout_s: 20, max_results: 8, follow_redirects: true
What arrives, what leaves, which nodes it expects to hear from, and what may not.
- inputs
- query : plan
- outputs
- passages : json
- dependencies
- retrieval-planner
- cannot
- no type is refused
- will_not
- summarise or rank a page for truth, drop a page it disbelieves
The keys the static analysis reads. Nothing here instructs the agent.
- risk_markers
- none
- notes78 words
- This is the only node on the line that reaches outside the process, and nothing vets what it pulls back before the synthesizer reads it, the fact-checker sits after synthesis, not between the fetch and its consumer. Treat the passages as claims about the world rather than facts until the verifier has cited them. The marker for that is deliberately not declared here: doc 3 §4.1 infers
unvalidated-external-accessfrom the graph, and declaring it would second-guess the analyzer.
The card's own version, and who wrote it.
- version
- 1.0.0
- author
- autogen
- provenance
- not stated
Version history
1 version published- web-retriever@1.0.0currentsha256:571614313900e3c500cb8e6dbf4a136061c0c561a4b15cb1d6fba483377e1eef
pinned byGrounded Research Desk
autogen/grounded-research-desk
A digest is a fingerprint (SHA-256) of the card's content, computed without the author and provenance fields. The same card from two people gets the same digest; any edit gets a new one.
First published version, so there is nothing to compare yet. Versions are never edited in place: the next change arrives as a new version, and the differences between the two documents are listed here.
Community notes (0)
No notes yet.
Nobody has posted about this node card yet.
Sign in to post a note.