zeph_config/
knowledge.rs

1// SPDX-FileCopyrightText: 2026 Andrei G <bug-ops>
2// SPDX-License-Identifier: MIT OR Apache-2.0
3
4//! Knowledge-ingest configuration (`[knowledge]`).
5//!
6//! This module defines [`KnowledgeConfig`], the optional TOML section that controls
7//! the `zeph knowledge ingest` command (spec-067, Phase 1). All fields carry sane
8//! defaults so an existing config that omits `[knowledge]` entirely still works.
9//!
10//! # Configuration example
11//!
12//! ```toml
13//! [knowledge]
14//! ingest_provider = "fast"   # from [[llm.providers]]; Phase 2 graph extraction
15//! concurrency = 3
16//! max_documents = 0          # 0 = unlimited
17//! recall_include_imported = true
18//! transcript_scope = "current-project"
19//! ```
20
21use serde::{Deserialize, Serialize};
22
23/// Configuration for the `zeph knowledge` subsystem (spec-067 Phase 1).
24///
25/// Deserialised from the optional `[knowledge]` table in `config.toml`.
26/// Missing fields fall back to [`Default`] values — no migration required for
27/// existing configs that omit the section entirely.
28#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
29#[serde(default)]
30pub struct KnowledgeConfig {
31    /// Provider name from `[[llm.providers]]` used for graph extraction (Phase 2).
32    ///
33    /// Empty string → fall back to `[memory.graph].extract_provider` → primary provider.
34    /// The notes-sink (Phase 1) does not perform LLM calls; this field is reserved for
35    /// Phase 2 graph extraction and ignored until then.
36    pub ingest_provider: String,
37
38    /// Maximum number of concurrent document-processing tasks during batch extraction (Phase 2).
39    ///
40    /// Has no effect in Phase 1 (notes sink processes files sequentially via
41    /// `IngestionPipeline`). Stored now for config stability.
42    pub concurrency: usize,
43
44    /// Maximum number of documents processed per `ingest` run; `0` = unlimited.
45    ///
46    /// The CLI `--max-documents` flag overrides this value when non-zero.
47    /// Bounds cost per run (spec-067 NFR-002).
48    pub max_documents: usize,
49
50    /// Whether semantic-recall results include rows imported via `zeph knowledge ingest`.
51    ///
52    /// When `false`, only conversation-derived memory rows are surfaced.
53    /// Phase 2 recall integration honours this flag; Phase 1 notes-sink writes use it
54    /// only to document intent.
55    pub recall_include_imported: bool,
56
57    /// Scope of transcript sources eligible for ingest.
58    ///
59    /// Only `"current-project"` is honoured in Phase 1 (INV-6: project-root-anchored
60    /// sources only). Other values are reserved for future cross-project modes.
61    pub transcript_scope: String,
62}
63
64impl Default for KnowledgeConfig {
65    fn default() -> Self {
66        Self {
67            ingest_provider: String::new(),
68            concurrency: 3,
69            max_documents: 0,
70            recall_include_imported: true,
71            transcript_scope: "current-project".to_owned(),
72        }
73    }
74}
zeph_config/knowledge.rs

zeph_config/
knowledge.rs
