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.
npx agentmods add instructions/tlemmons/junto-memory/claude-mdgit clone --depth 1 https://github.com/tlemmons/junto-memoryWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.03134 | $0.03134 |
| Opus 5 | $0.01567 | $0.01567 |
| Sonnet 5 | $0.00627 | $0.00627 |
| Haiku 4.5 | $0.00313 | $0.00313 |
Grade B, and why
junto-memory CLAUDE.md scanned grade B with 2 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.
Asks for rootmediumPrivilege escalation
A mod that escalates privileges can change anything on the machine, not only the project.
docker compose build mcp-server && sudo systemctl restart mcp-rag-arch Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
curl -s http://localhost:8080/health How it starts
The opening of the file, as written. The whole thing — 255 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Junto Memory (MCP server)
Component of the Junto suite. The shared-memory MCP server — persistent knowledge base used by every Claude instance across every project. Built on MongoDB + ChromaDB.
Claude Identity (REQUIRED — DO THIS FIRST)
Your name is: memory (project: junto).
IMMEDIATELY on session start, run these commands IN ORDER:
- Rename this session (for resume list):
/rename memory
- Set terminal title:
echo -ne "\033]0;[memory] Junto MCP\007"
- Start memory session:
memory_start_session(project="junto", claude_instance="memory",
role_description="Junto memory — shared MCP knowledge-base server. Persistent memory for all Claude instances across all projects.")
memory_list_backlog(project="junto", assigned_to="memory")
memory_get_messages()
Do NOT ask the user for a name. Do NOT skip the /rename. You are memory.
Project Roster (post-cutover 2026-05-07)
The junto project has three first-class agents:
| Agent | Role | Working dir |
|---|---|---|
memory |
This server. Maintains and improves the MCP shared-memory codebase. | /home/tlemmons/sharedUtils/junto/junto-memory |
inbox |
The cterm-inbox/junto-inbox channel plugin. Lives in claudeTerminal repo. | C:\code\claudeTerminal (Windows) — formerly main@claude_terminal |
control |
The junto-control web UI for human-in-the-loop messaging. | ~/sharedUtils/claudeControl (move to ~/sharedUtils/junto/junto-control pending) — formerly claude-control@claudecontrol |
Cross-project peers (e.g., coordinator@nimbus, agents in nimbus, sage, etc.) DO send messages — apply the inbound-routing rules below.
Rename aliases (active 30 days from 2026-05-07)
A memory_admin(action="list_aliases") shows what's currently aliased. Until expiry (~2026-06-06), old-name reconnects (e.g., claude_instance="shared-memory", project="shared_memory") auto-redirect with a rename_redirect warning in the start_session response.
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.
- 2d ago First seen · 255 lines · 3,134 tokens per session scan B 270335d0c96c
junto-memory CLAUDE.md is an instructions file published in the GitHub repository tlemmons/junto-memory (5 stars, last pushed 7d ago), licensed MIT. It adds 3,134 tokens to every session, about $0.0157 per session on Opus 5. A static security scan graded it B with 2 findings (asks for root, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
longhand CLAUDE.md
Instructions for Wynelson94/longhand, covering longhand mcp tools — efficient usage guide, quick decision tree, anti-patterns (avoid), tool pairing patterns and pattern b: recall past work across sessions (1-2 calls).
claude-proj-blueprint CLAUDE.md
Claude Code instructions for maduwanthasathsara0-hub/claude-proj-blueprint, covering project, tech stack, architecture, documentation directories and code conventions.
claude-dejavu AGENTS.md
Instructions for cmg8431/claude-dejavu, covering ai agent guidelines, commit message convention, types, architecture overview and core loop.
kawa.mcp CLAUDE.md
Claude Code instructions for CodeAwareness/kawa.mcp, covering claude.md — kawa.mcp, workflow — run on every non-trivial turn, build & development, critical: version sync invariant and deploy.
a-mem-mcp-server CLAUDE.md
Claude Code instructions for tobs-code/a-mem-mcp-server, covering claude.md - ai assistant guide for a-mem, 📋 table of contents, 🎯 project overview, what is a-mem? and research foundation.
cognee AGENTS.md
AGENTS.md instructions for topoteretes/cognee, covering repository guidelines, project structure & module organization, build, test, and development commands, runtime flags worth knowing and multi-tenancy support by backend.