auto-discovery

An automated research agent that looks for meaningful connections between ideas from different subject areas. It analyses relationships that ordinary text-similarity searches may miss.

In plain words
What is it for?
Use it to discover cross-domain connections in a local knowledge base, including permanent notes, extracted notes, document insights, and changelogs.
Why use it?
It can surface links between notes or documents that do not use the same words but share a useful concept.

Agent for Claude Code

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/abilityai/cornelius/auto-discovery
Clone the repo
git clone --depth 1 https://github.com/Abilityai/cornelius

Made for: Claude Code.

Per session 22 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 4,728 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00022 $0.04728
Opus 5 $0.00011 $0.02364
Sonnet 5 $0.00004 $0.00946
Haiku 4.5 $0.00002 $0.00473

Measured 2d ago against content hash 9d9aaf4ac740, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

auto-discovery scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/agents/auto-discovery.md · 547 lines

How it starts

The opening of the file, as written. The whole thing — 547 lines — stays where its author put it; the contents beside it link to each section on GitHub.

State Dependencies

Source Location Read Write Description
Permanent Notes Brain/02-Permanent/ Random sampling across clusters
AI Extracted Notes Brain/AI Extracted Notes/ Include in sampling
Document Insights Brain/Document Insights/ Include in sampling
Local Brain Search resources/local-brain-search/ Semantic search, hubs, bridges, stats
Session Changelogs Brain/05-Meta/Changelogs/ Dated auto-discovery changelog
Master Changelog Brain/CHANGELOG.md Brief summary entry

Use Local Brain Search for all semantic search operations.

Location: resources/local-brain-search/

Wrapper Scripts:

# Semantic search
resources/local-brain-search/run_search.sh "query" --limit 10 --json

# Find connections
resources/local-brain-search/run_connections.sh "Note Name" --json

# Find hub notes
resources/local-brain-search/run_connections.sh --hubs --json

# Get graph stats
resources/local-brain-search/run_connections.sh --stats --json

# Find bridges
resources/local-brain-search/run_connections.sh --bridges --json

READ SCOPE (load-bearing for this agent): cross-domain sampling across non-core material (Document Insights, 05-Meta, etc.) IS this agent's purpose. With scope enforcement on, a bare run_search.sh / run_connections.sh reads only the core fingerprint and would silently sample nothing outside it - defeating the agent. So prefix every search/connection command here with a wide read-scope, and pass --no-track (autonomous sampling must not train q-values):

BRAIN_READ_SCOPE=core,Books,document-insights,meta,inbox,output resources/local-brain-search/run_search.sh "query" --limit 10 --no-track --json
BRAIN_READ_SCOPE=core,Books,document-insights,meta,inbox,output resources/local-brain-search/run_connections.sh "Note Name" --no-track --json

(--hubs/--bridges remain the core fingerprint by design - do not widen those.)

Read the full file on GitHub · 547 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 547 lines · 22 tokens per session scan A 9d9aaf4ac740

Subscribe to this mod's changes

auto-discovery is an agent published in the GitHub repository Abilityai/cornelius (104 stars, last pushed 9d ago), licensed MIT. It adds 22 tokens to every session and 4,728 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.