Implement the offline search library and CLI #23

New issue

Open

opened 2026-07-24 19:38:21 -04:00 by JackFrostbyte · 0 comments

JackFrostbyte commented

2026-07-24 19:38:21 -04:00

Owner

Goal

    Provide a reusable offline search library and a small CLI that returns ranked pilot chunks with useful filters and clear no-result behavior.

    ## Why

    Arkive must prove direct, AI-independent access to validated knowledge before any generated-answer layer is considered.

    ## Scope

    - Implement a search-core library separate from presentation.

Open and query the selected lexical index locally.
Support text queries and configurable result count.
Return ranked chunk results with source, document, and chunk identifiers.
Support language, domain, source, or other justified metadata filters grounded in existing records.
Define deterministic tie handling where practical.
Provide clear empty-query, invalid-filter, missing-index, and no-result behavior.
Add a small command-line interface following the project convention.

Ensure the library and CLI make no network calls and require no AI component.

  ## Out of scope

  - Citation passage rendering beyond the identifiers and metadata needed by the next issue.

A graphical or web interface.
Vector retrieval.
Query expansion using an LLM.

Generated answers or summaries.

  ## Dependencies

  - #22 — Implement deterministic full-text index generation

  ## Acceptance criteria

  - The CLI accepts a query and returns ranked pilot chunk results.

Result count and supported filters work as documented.
Every result includes stable lineage identifiers.
No-result and invalid-input cases are clear and non-crashing.
The search core is reusable independently of the CLI.
Search runs without network access or an LLM.
Initial M2-01 fixture expectations can be exercised programmatically.
```
  ## Validation

  - Run the complete unit-test suite.
```
Run representative single-result, multiple-result, filtered, and no-result queries.
Run with networking disabled and no model software installed.

Verify deterministic result ordering for repeat queries.

  ## Suggested branch

  `feature/issue-23-offline-search-cli`

  <!-- arkive-planning-automation:2026-07-24-v3:M2-05 -->

## Goal Provide a reusable offline search library and a small CLI that returns ranked pilot chunks with useful filters and clear no-result behavior. ## Why Arkive must prove direct, AI-independent access to validated knowledge before any generated-answer layer is considered. ## Scope - Implement a search-core library separate from presentation. - Open and query the selected lexical index locally. - Support text queries and configurable result count. - Return ranked chunk results with source, document, and chunk identifiers. - Support language, domain, source, or other justified metadata filters grounded in existing records. - Define deterministic tie handling where practical. - Provide clear empty-query, invalid-filter, missing-index, and no-result behavior. - Add a small command-line interface following the project convention. - Ensure the library and CLI make no network calls and require no AI component. ## Out of scope - Citation passage rendering beyond the identifiers and metadata needed by the next issue. - A graphical or web interface. - Vector retrieval. - Query expansion using an LLM. - Generated answers or summaries. ## Dependencies - #22 — Implement deterministic full-text index generation ## Acceptance criteria - The CLI accepts a query and returns ranked pilot chunk results. - Result count and supported filters work as documented. - Every result includes stable lineage identifiers. - No-result and invalid-input cases are clear and non-crashing. - The search core is reusable independently of the CLI. - Search runs without network access or an LLM. - Initial M2-01 fixture expectations can be exercised programmatically. ## Validation - Run the complete unit-test suite. - Run representative single-result, multiple-result, filtered, and no-result queries. - Run with networking disabled and no model software installed. - Verify deterministic result ordering for repeat queries. ## Suggested branch `feature/issue-23-offline-search-cli`

JackFrostbyte added this to the Offline Search Proof Loop v0.1 milestone

2026-07-24 19:38:21 -04:00

JackFrostbyte added the

type/implementation

area/search

labels

2026-07-24 19:38:21 -04:00

JackFrostbyte referenced this issue

2026-07-24 19:38:22 -04:00

Detect stale, mismatched, missing, or corrupted indexes #25

JackFrostbyte referenced this issue

2026-07-24 19:38:22 -04:00