quarry AGENTS.md

quarry AGENTS.md is an instructions file for Codex, OpenCode from punt-labs/quarry. It costs 1,961 tokens per session, scanned A, original, MIT.

Project instructions for Quarry, a local tool that stores documents and conversation fragments so an AI agent can search them by meaning across sessions.

In plain words
What is it for?
Use it to ingest, search, recall, list, or delete project knowledge through agent tools, shortcuts, and automatic hooks.
Why use it?
It provides memory outside the agent’s limited conversation window, including support for many document formats and scanned files.

Instructions file for CodexOpenCode

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 instructions/punt-labs/quarry/agents-md
Clone the repo
git clone --depth 1 https://github.com/punt-labs/quarry

Made for: Codex, OpenCode.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for quarry AGENTS.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/punt-labs/quarry/agents-md.svg)](https://agentmods.dev/instructions/punt-labs/quarry/agents-md)
Your own site
<a href="https://agentmods.dev/instructions/punt-labs/quarry/agents-md"><img src="https://agentmods.dev/badge/instructions/punt-labs/quarry/agents-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 1,961 This file is loaded in full into every session.
When invoked 1,961 The same file — it is already loaded in full.
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 $0.01961 $0.01961
Opus 5 $0.00981 $0.00981
Sonnet 5 $0.00392 $0.00392
Haiku 4.5 $0.00196 $0.00196

Measured 4d ago against content hash 6b876cbd6b91, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

quarry AGENTS.md 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 4d 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.

AGENTS.md · 173 lines

How it starts

The opening of the file, as written. The whole thing — 173 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Quarry for Agents

Quarry gives AI agents persistent semantic memory. Index documents, search by meaning, and recall knowledge across sessions — all local, no API keys.

What Quarry Does

You have a context window. It fills up and gets compacted. Quarry sits outside that window as durable storage: you ingest documents and conversation fragments, then search them later by meaning. A query about "margins" finds passages about profitability even if they never use that word.

Quarry handles 20+ formats natively: PDFs (with OCR for scanned pages), source code (AST-aware splitting), spreadsheets, presentations, HTML, Markdown, LaTeX, DOCX, and images.

Integration Model

Quarry integrates with Claude Code as a plugin that provides three layers:

  1. MCP tools — the core interface. Search, ingest, list, delete, manage databases.
  2. Slash commands — user-facing shortcuts that call MCP tools with formatting logic.
  3. Hooks — automatic background behaviors that run without user intervention.

MCP Tools

These are the tools you call directly. Read-only tools are synchronous; write tools are fire-and-forget (return immediately, process in background).

Tool Purpose Execution
find Hybrid search (vector + BM25 via RRF) with optional filters (collection, document_filter, page_type, source_format, agent_handle, memory_type) Sync
show Retrieve document metadata or a specific page's full text Sync
list List documents, collections, databases, or registrations Sync
status Database stats: doc/chunk counts, size, model info Sync
ingest Ingest a file or URL (auto-discovers sitemaps) (supports agent_handle for expertise seeding) Background
remember Ingest inline text content with a document name (supports agent_handle, memory_type, summary for memory tagging) Background
delete Delete a document or collection Background
register_directory Register a directory for incremental sync Background
deregister_directory Remove a directory registration Background
sync_all_registrations Sync all registered directories Background
use Switch the active named database Sync

Read the full file on GitHub · 173 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. 4d ago First seen · 173 lines · 1,961 tokens per session scan A 6b876cbd6b91

Subscribe to this mod's changes

quarry AGENTS.md is an instructions file published in the GitHub repository punt-labs/quarry (3 stars, last pushed 4d ago), licensed MIT. It adds 1,961 tokens to every session, about $0.0098 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 2026-08-31.