Command documentation

inspect

Prove claims by inspecting exact objects and source-backed rows.

Purpose

This command is available over HTTP, MCP and the local CLI. HTTP responses use the compact public response shape with summaries, warnings and bounded tables.

MCP connection

Use endpoint https://bible.grod.es/mcp with tool bible_inspect. The MCP Server Card is /.well-known/mcp/server-card.json.

Parameters

Name	Required	Description
`target`	yes	Exact object to prove with source-backed rows: reference, Strong key, resource, variant, entity, topic, source, or object ID. Target grammar: natural Bible reference, or prefix:value. Examples: Deut 32:8, John 1, ref:Deut 32:8, strong:H7225, entity:Paul, source:LXX, variant:<id>, resource:<id>.
`limit`	no	Maximum proof rows per inspection table or lens. Inspect is for verification, so the default is higher than synthesis commands, but still bounded to avoid turning one request into a full database dump. Default 100, max 500.
`depth`	no	Expansion depth from 0 to 10. 0 returns direct facts for the target; 1 is the default useful context; higher values include more related context and are clamped by the server --max-depth instead of failing. For inspect it expands related proof rows and raw/provenance context where requested.
`kind`	no	Inspection lens/table family, such as text, apparatus, variants, witnesses, morphology, resources, entities, or lineage. It decides which proof rows are returned.
`include`	no	Optional proof expansions to include, such as related rows. Use raw for raw source rows and provenance for source lineage.
`provenance`	no	Controls provenance detail independently from domain includes: none omits extra provenance sections, summary returns compact source lineage, and full returns the most detailed provenance available for the command.
`raw`	no	Inspect-only boolean that adds raw source rows beside normalized inspection output. Use it for verification, not normal synthesis.
`resource_type`	no	Resource type filter for resource/search/notable/source inventory rows, such as commentary, dictionary, article, dataset, or media.
`min_confidence_score`	no	Minimum confidence threshold from 0 to 1. It filters numeric confidence_score values, or documented label-derived scores high=1, medium=0.6, low=0.3; rows without numeric confidence are not removed, and basis-only values such as source_attested or algorithmic are not treated as confidence scores.
`review_status`	no	Review-status filter for source-backed rows. Use any to include all rows, or a specific status to separate reviewed, seeded, pending, or diagnostic material.

Required option sets: [["target"]]

Examples

HTTP

curl -s 'https://bible.grod.es/api/inspect?kind=apparatus&provenance=full&raw=true&review_status=any&target=John+7%3A53'

MCP

{
  "arguments": {
    "kind": "apparatus",
    "provenance": "full",
    "raw": true,
    "review_status": "any",
    "target": "John 7:53"
  },
  "name": "bible_inspect"
}

CLI

bible inspect --kind apparatus --provenance full --raw --review-status any John 7:53

Caveats for this command

Textual variants and witness signals are source-backed but coverage varies by passage and tradition.
DSS material is fragmentary.
LXX GreekResources data is lemma-level and not a full critical apparatus.
The runtime does not call LLM or embedding APIs.