Skip to content
EverrunsEverruns

Retrieval Citations

IDcitation_retrieval
CategoryKnowledge
Featurescitations
DependenciesNone

Turn the sources an agent retrieves into claim-level provenance on its answer. When the agent grounds a reply in a knowledge search, citation_retrieval links each grounded sentence to the passage that backs it, and the UI renders those links as inline numbered chips with a hover preview and a deduped Sources strip.

It contributes no tools and no system prompt, and it never rewrites the model’s answer. After the agent responds, it inspects the turn’s retrieval results, aligns each retrieved passage to the sentence it best supports, and attaches a citation there. Alignment is deterministic token overlap — no extra model call — so it is model- and provider-agnostic.

How it looks

Section titled “How it looks”

Grounded sentences get an inline numbered chip. Hovering a chip previews the source (title, snippet, and link); a deduped Sources strip sits below the message. When Citation Verification is also enabled, each source carries a faithfulness badge.

An assistant message with inline numbered citation chips on grounded sentences, a hover popover previewing the cited source's title, snippet, and URL, and a "Sources" strip below the message listing three deduped sources with verified / unsupported / unverified badges.

Feed

Section titled “Feed”

Reads the results of the agent’s knowledge retrieval tools:

ToolSource shape
search_indexKnowledge-index chunks (kchk_…) — source_uri, document_title, snippet, location
search_knowledgeKnowledge-base entries (kbe_…) — resource, title, snippet

Both shapes are normalized into the shared citation envelope. A retrieval result with no usable snippet is skipped — there is nothing to align a claim to.

This capability is a no-op on its own. It cites what the agent retrieves, so it only produces citations when the agent also has a retrieval capability (a knowledge index or knowledge base) that surfaces search_index / search_knowledge results. With no retrieval feed, there is nothing to cite and the answer is unchanged.

Configuration

Section titled “Configuration”
FieldTypeDefaultDescription
min_overlapnumber (0–1)0.5Minimum token-overlap ratio for a retrieved passage to attach to a sentence. Overlap is shared tokens / passage tokens, so 0.5 means at least half the passage’s distinctive words appear in the sentence.

Raise min_overlap for stricter, higher-precision attachment (fewer chips, each more defensible); lower it for broader coverage when the model paraphrases retrieved text.

Notes

Section titled “Notes”
  • No answer rewrite: the streamed answer is never changed; annotations attach to sentence spans after generation.
  • Alignment is lexical: a sentence that paraphrases a source heavily enough to fall below min_overlap will not be cited. This favors precision over recall.
  • Enabled by default: citation_retrieval is part of the generic (default) harness, so any agent with a retrieval feed gets citations automatically.
  • Persistence: annotations ride the message in the event log, so they survive reload, forking, and session export — no separate store.
  • Org scoping: sources are derived only from already-authorized retrieval results, so a citation can never reference a document the requesting org cannot read.

See Also

Section titled “See Also”