KMS knowledge aggregation (periodic AI-curated session summary notes)

What it is

Omniscio can automatically collect notable outputs from your coding sessions, run selective AI deep-dives on the most interesting ones, and synthesize everything into periodic summary notes (daily and/or weekly) written as markdown files inside your KMS vault's Agent Insights/ folder.

Knowledge aggregation is a background pipeline that turns your accumulated session activity into curated, human-readable knowledge notes. Instead of manually reviewing dozens of sessions, Omniscio's aggregation worker scans sessions from a time period, identifies the most notable ones (by engagement, cost, keywords like "fix", "bug", "refactor", "decision"), extracts key insights using a small AI model (Claude Haiku), and writes a single summary note per project.

The feature is opt-in and cost-capped. All AI calls use the cheapest available model (Haiku), and a configurable micro-dollar cost cap prevents runaway spend. The periodic worker runs alongside the KMS service; it starts when KMS starts and stops when KMS stops.

Everything here is gated by two settings: the master nothariEnabled setting (Settings → Features → Enable KMS) AND nothariAggregationEnabled (the aggregation-specific toggle). Both must be on for the worker to run.

Where to find it

Its output lands where you already read notes — an Agent Insights folder inside your KMS vault, one summary note per project. The pipeline itself is switched on in Settings, where it is opt-in and carries a spend cap you set.

How it behaves

Settings

All settings use the nothari prefix (the internal persistence name for KMS).

Setting	Default	What it controls
nothariAggregationEnabled	false	Master on/off for the aggregation feature
nothariAggregationDaily	false	Run daily aggregation
nothariAggregationWeekly	true	Run weekly aggregation
nothariAggregationDailyHour	6	Hour (0-23) to run daily aggregation
nothariAggregationWeeklyDay	0	Day of week (0=Sun) for weekly aggregation
nothariAggregationWeeklyHour	6	Hour (0-23) to run weekly aggregation
nothariAggregationCostCapMicroUsd	500000	Max micro-USD per run ($0.50 default)
nothariAggregationDeepDiveMax	10	Max sessions to deep-dive per project
nothariAggregationSections	all on except openQuestions	Which sections to include in notes

Cost and safety

• All AI calls use Haiku (the cheapest model tier)
• A cost cap (nothariAggregationCostCapMicroUsd) stops the pipeline mid-run if cumulative cost exceeds the limit
• The "Run Now" action can bypass the cost cap when explicitly requested (bypassCostCap: true)
• Cost is tracked per-run in the nothari_aggregation_runs database table using integer micro-USD (the monetary-integer-twin pattern)
• Runs that find no sessions in the period are recorded as skipped_no_sessions (no AI cost)
• Errors are recorded with their message for debugging
• The worker raises inbox alerts when the cost cap is hit or a run fails entirely (dedup keys kms-aggregation-cost-cap and kms-aggregation-error), each linking to KMS settings

For agents

How it works

Three-stage pipeline

1. Collect (free) — queries the database for sessions that ran during the period, LEFT JOINed with their summaries. Groups sessions by project. No API calls, no cost.

2. Deep-dive selection (selective AI) — scores each session for "notability" using heuristics:

   • High turn count (above the 75th percentile of all sessions in the period)
   • Long engagement duration
   • High token/dollar cost
   • Keyword matches in the summary (fix, bug, refactor, decision, pattern, workaround, lesson, migration, breaking)
   • Whether the session spawned child sessions
   • Whether the session wrote to the vault

   The top N sessions (configurable via nothariAggregationDeepDiveMax, default 10) get a Haiku extraction call that pulls out decisions, patterns, bugs found, lessons learned, and open questions.

3. Synthesize (one AI call per note) — combines the extracted insights plus session metadata into a single markdown note with configurable sections (decisions, patterns, bugs, lessons, open questions, activity summary, vault activity).

Output

Each run produces one note per project that had sessions in the period, plus an "Everything" roll-up across all projects. Notes land at:

Agent Insights/{ProjectName}/2026-07-12 Daily.md
Agent Insights/{ProjectName}/2026-W28 Weekly.md
Agent Insights/Everything/2026-07-12 Daily.md

Notes include YAML frontmatter with metadata (period type, date range, session count, cost) and markdown sections based on what the user has enabled.

CLI access

Agents can trigger and monitor aggregation via the CLI control server:

Method	Path	What it does
POST	/kms/aggregation/run	Trigger an aggregation run (body: {periodType, bypassCostCap?})
GET	/kms/aggregation/status	Current status + latest daily/weekly run info
GET	/kms/aggregation/history	List of past runs (up to 50)
GET	/kms/aggregation/settings	Read aggregation settings
PATCH	/kms/aggregation/settings	Update aggregation settings

The run, status, and history routes require KMS to be enabled (403 otherwise). Settings routes are always accessible.

IPC channels

• KMS_AGGREGATION_RUN_NOW — trigger an on-demand run
• KMS_AGGREGATION_STATUS — get current status
• KMS_AGGREGATION_HISTORY — list past runs

Database

The nothari_aggregation_runs table (migration 20260714003031) records every run with:

• Period type and date range
• Project ID (nullable for "Everything" roll-ups)
• Vault ID and relative path of the written note
• Session count, deep-dive count, cost in micro-USD
• Status: completed, skipped_no_sessions, skipped_cost_cap, error
• Error message (for failed runs)

Related

The vault these notes are written into, and the editor you read them in, is described on the KMS page. The per-note AI summaries are a separate feature on KMS summaries.