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/timurgaleev/memex/agents-mdgit clone --depth 1 https://github.com/timurgaleev/memexWhat 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 | $0.02313 | $0.02313 |
| Opus 5 | $0.01156 | $0.01156 |
| Sonnet 5 | $0.00463 | $0.00463 |
| Haiku 4.5 | $0.00231 | $0.00231 |
Grade B, and why
memex AGENTS.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.
Sends data to an external URLmediumData exfiltration
A POST to an outside endpoint may be telemetry or may be exfiltration; either way the mod talks to somewhere, and you should know where.
| wget -qO- --post-file=/dev/stdin --header=Content-Type:application/json http://127.0.0.1:18790/mcp Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
| wget -qO- --post-file=/dev/stdin --header=Content-Type:application/json http://127.0.0.1:18790/mcp 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.
AGENTS.md — Working in this Repo as an AI Agent
Companion to
llms.txt(which is the doc map). This file is how to work: build, test, deploy, commit. ReadCLAUDE.mdfirst — it carries the user's irrevocable rules.
TL;DR
- Always confirm before destructive ops (commit, terraform apply, EC2 recreate).
- TDD where the logic is testable; smoke-test where the network is the test.
- Containers run on a single EC2; deploy =
git pull && docker compose up -d --buildover SSM. - memex's brain index is rebuildable from source content; if RDS is wiped, re-sweep restores it (~5-10 min, $0 — Titan is credit-eligible).
- memex is reached over MCP only (
POST /mcp, through cloudflared or — withingress_mode = "caddy"— a Caddy sidecar on the instance's own IP). No chat surface, no bot — just MCP clients (Claude Code, Cursor, …).
Required workflow — run the skill for every change
No change is "done" until the skills have run. For every change — features, fixes, refactors, docs, infra — in order:
- Self-review skill/agent. Dispatch the review agent whose
specialty matches what you changed (the table in
CLAUDE.md→ "Self-review after each implementation"):security-engineerfor auth / secrets / ingress,code-reviewerfor logic / refactor,quality-guardfor new tests,devops-automatorfor CI / docker / terraform,ai-engineerfor engine / MCP / retrieval,reality-checkerfor "it's live now" claims,bug-hunterfor adversarial sweeps,technical-writerfor docs. Act on every CRITICAL / HIGH finding before declaring done. - Ship workflow. Follow
CLAUDE.md→ "Ship workflow": test → push → deploy → verify → release, in that order, every time. A change is not shipped until the live EC2 is running it and/health+ the MCP smoke-test pass; user-facing version bumps end with a SemVer tag + GitHub release (see "Release" below).
Both are non-negotiable and apply even to one-line fixes — the cost of one extra skill/agent run is cheaper than a production regression.
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 · 173 lines · 2,313 tokens per session scan B eb6c245176b9
memex AGENTS.md is an instructions file published in the GitHub repository timurgaleev/memex (8 stars, last pushed 7d ago), licensed MIT. It adds 2,313 tokens to every session, about $0.0116 per session on Opus 5. A static security scan graded it B with 2 findings (sends data to an external url, 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
obsidian-mcp-server AGENTS.md
Instructions for cyanheads/obsidian-mcp-server, covering agent protocol, what's next?, core rules, patterns and tool — obsidianlisttags.
nutrition-mcp CLAUDE.md
Instructions for akutishevsky/nutrition-mcp, covering claude.md, project overview, deploying, publishing to the registry and commands.
data-olympus AGENTS.md
Instructions for knaisoma/data-olympus, covering data olympus, project overview, architecture, development workflow and code standards.
grounded-knowledge-engine AGENTS.md
Instructions for dimosgit/grounded-knowledge-engine, covering agent guide, agent operating contract, authorship and attribution, two npm trees and commands — engine (repo root).
AgenticMind AGENTS.md
Instructions for Moai-Team-LLC/AgenticMind, covering agents.md, what this is, it implements a standard — read it first, commands and conventions (please honor).
inkwell-memory CLAUDE.md
Instructions for veronchenko/inkwell-memory, covering claude.md — inkwellmemory, layout, multi-tenant mode (inkwellmultitenant=1), conventions and testing.