how nectar compares

Page: compare-nectar | Slug: compare/nectar | Version: 1.0 | Date: July 2026 | Status: Draft for approval

The nectar comparison hub. Not in the primary nav; reached from footer links and in-page buttons. Honest, sourced, favors nectar without pretending competitors are weak. Brand voice: lowercase wordmarks, sentence case, mono for ids, paths, and commands, no emoji.

---

#Hero

how nectar compares

nectar did not invent codebase search. plenty of good tools give files identity, describe code with an LLM, or index a repo semantically. what nectar does that no other tool does is combine all of it into one thing that composes with your agent's memory. here is an honest, sourced look at the field, competitor by competitor.

• button: get the apiary → /install

#The honest claim

we do not claim nectar invented semantic code search. the defensible claim is narrower and it is this: nectar is the first system to combine daemon-minted file identity, plain-language file description, a durable store you own, and recall that fuses with your conversation memory, all in a background service that plugs into the assistants you already use. each of those pieces exists somewhere in prior work. the combination does not.

#The five things nectar puts together

1. a stable file identity minted by the daemon, not derived from the file's contents or its path
2. a plain-language description of each file, written by a language model
3. a durable store you own that does both exact lookups and meaning-based search
4. recall that fuses code with your session and skill memory in a single query
5. a portable, committed map so a teammate inherits the understanding for free

the closest single tool covers two of these. none covers more than two.

#The field at a glance

tool	stable identity	plain-language description	meaning-based store	watches files	survives a rename
nectar	yes, minted	per file	yes, deep lake	yes	yes
openwiki	no	per wiki page	no index	on a schedule	no
smith	no, path-keyed	per file	sidecar files	manual	no
aura	yes, per function	no	shadow branch	git-hooked	yes
coderag and the AST-chunk family	no	per code chunk	vector db	yes	no

read it this way: each column is covered by some tool, but no other tool fills the whole row.

#The head-to-heads

• nectar vs openwiki: the free committed-wiki approach, and why "the agent greps a doc" is not the same as "the agent recalls the right file."
• nectar vs smith: the closest prior art for describing files, and the identity and storage gaps nectar fills.
• nectar vs aura: the clearest predecessor for stable identity, at function granularity, and how nectar applies the same idea to memory.
• nectar vs coderag: the dominant AST-chunk pattern, and the deliberate choice to describe files instead of symbols.

#What we credit

nectar is a synthesis, and we say so. the identity split is aura's idea. the commit-your-descriptions idea is smith's. delta indexing by content hash is grove's and cartog's. semantic search over embeddings is the coderag family's. we borrow openly and name every source. the originality is in the composition, not any single piece.

#How we keep this fair

every competitor fact on the head-to-head pages carries its source and the month we captured it. where we were not certain of a fact, we say so rather than overclaiming. these are honest comparisons that happen to favor nectar because the composition is genuinely different, not takedowns.

• button: get the apiary → /install
• button: read how nectar works → /nectar