Birth-session clustering + the prober graph-tools data seam (wave 7 chunk A)
Embeddings are dying (chunk B removes the machinery); the clusters they
used to carry become the session that created them — derived on read.
- New src/core/session-clusters.ts: birthSessionPartition groups merged
nodes by their FIRST SURVIVING provenance point's chatId (owner-locked
2026-09-23). Deletion is self-healing by construction: stripped-birth
survivors re-attribute to their earliest remaining contributor, and
empty-provenance nodes stay unclustered. Notes imports are birth-sessions
too (owner ruling) — their label is the file path.
- graphSnapshot: clusters = the birth partition filtered >= MIN_CLUSTER_SIZE
(the hull presentation rule), id = session id, label = session name (chat
title / file path; a deleted record renders unlabeled). The persisted
cluster-labelings join dies; no LLM naming pass needed anymore.
- settleGraphLayout derives the same partition from readNodes — drops its
readEmbeddings dep. One cluster concept everywhere.
- Prober data seam (core/graph.ts, pure ports-driven like the snapshot):
graphSessionIndex (every birth session: id, name, chat|file kind,
createdAt/updatedAt, birth-proposition count, recency-ordered) and
graphSessionPropositions (a session's birth propositions with the
extractor's latest understanding level and THAT event's own timestamp —
owner ruling: quiz outcomes are ledger evidence, they never move the
evaluation date). Teacher's tool handlers consume these in-process.
- New agent.graphTools subsection { enabled?: boolean } (owner ruling:
the toggle gates prober tool REGISTRATION — agent capability, agent
section, its own subsection, NOT the tool-toggles map; absent = ON).
Lenient sanitizer, strict settings-body parser, passthrough resolution,
and graphToolsEnabled(effective) = graph.enabled && toggle for Teacher's
registry wiring.
Gates: tsc clean, eslint clean, 923/923 tests, next build compiles.
Live-verified on the worktree dev data: 10 nodes → 2 clusters labeled
with their chat titles; session index + propositions carry levels and
their own timestamps.
This commit is contained in:
@@ -200,6 +200,14 @@ export interface AgentSettings {
|
||||
* engine must still have its key, else search is honestly unavailable.
|
||||
* Absent + no key = unavailable (no silent fallback, no default). */
|
||||
search?: { engine?: SearchEngineId };
|
||||
/** Prober graph tools (wave 7, owner ruling 2026-09-23): read-only
|
||||
* knowledge-graph views for the prober agent — a session list +
|
||||
* per-session propositions with understanding levels. Its SUBJECT
|
||||
* is the graph but its MECHANISM is agent capability (tool
|
||||
* registration), so it lives here in its OWN subsection, not in the
|
||||
* tool-toggles map. Absent = enabled (the toggle discipline); the
|
||||
* effective gate composes graph.enabled AND this switch. */
|
||||
graphTools?: { enabled?: boolean };
|
||||
/** System-prompt override SWITCHES keyed by segment path (absent key =
|
||||
* off). Switch-on requires the override file to exist (enforced by the
|
||||
* route); the switch is explicit so users keep a file on disk but run
|
||||
@@ -419,6 +427,14 @@ function sanitizeAgent(raw: Record<string, unknown>): AgentSettings {
|
||||
const engine = (raw.search as Record<string, unknown>).engine;
|
||||
if (engine === "tavily" || engine === "brave") out.search = { engine };
|
||||
}
|
||||
// Prober graph tools (wave 7): its own agent-section subsection —
|
||||
// lenient read side (a bad shape drops the key and the default ON
|
||||
// resumes; the write side's parser is strict).
|
||||
if (raw.graphTools !== null && typeof raw.graphTools === "object" && !Array.isArray(raw.graphTools)) {
|
||||
const enabled = (raw.graphTools as Record<string, unknown>).enabled;
|
||||
if (enabled === undefined) out.graphTools = {};
|
||||
else if (typeof enabled === "boolean") out.graphTools = { enabled };
|
||||
}
|
||||
return out;
|
||||
}
|
||||
|
||||
|
||||
Reference in New Issue
Block a user