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/alp82/aistack/agents-mdgit clone --depth 1 https://github.com/alp82/aistackWhat 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 | $0.04908 | $0.04908 |
| Opus 5 | $0.02454 | $0.02454 |
| Sonnet 5 | $0.00982 | $0.00982 |
| Haiku 4.5 | $0.00491 | $0.00491 |
Grade A, and why
aistack 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 3d 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.
How it starts
The opening of the file, as written. The whole thing — 348 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AI Stack Webapp
This is a webapp for sharing AI Stacks so that users can compare and choose the best set of tools for their needs.
Development Process
- Web App is running on localhost:3019 - I am taking care of the dev servers running properly in my own terminal
- Convex is running already as well
- Schema changes are often required and okay
- Avoid backwards-compat, prefer convex migrations
Prod database access
Never point the local Convex CLI at prod. That broke the local setup multiple times. All prod operations run on the server over ssh ([email protected]). Two scripts wrap this:
# Any Convex CLI command against prod, executed on the server.
# The admin key is minted on the server per call and never stored here.
scripts/convex-prod.sh data models --limit 20
scripts/convex-prod.sh run migrations/<name>:run
# Replace the local dev database with a fresh prod export.
# Destructive locally (auth tables included - log in again). Prod is only read.
scripts/sync-prod-db.sh
The local database lags behind prod. Before work that depends on real rows (catalog data entry, migration dry runs), run scripts/sync-prod-db.sh first. It copies what prod HAS: for a reading no machine has published yet, see "Publishing a measured reading to the local backend".
Prod deploys run through GitHub Actions exclusively. A push to main triggers .github/workflows/deploy-convex.yml, which pulls the ~/aistack checkout on the server and runs pnpm convex deploy there. Never run npx convex deploy against prod from a local machine. To run a new migration on prod: push to main, wait for the workflow, then scripts/convex-prod.sh run migrations/<name>:run.
Never pass --push to scripts/convex-prod.sh, and never give it deploy, dev, or push. The CLI runs from a minimal project dir on the server that holds no convex/ directory, so anything that writes code pushes an EMPTY function set and prod loses every function and every index at once. That is a full outage, and it happened on 2026-08-24. The script now refuses those four forms, and the recovery is the workflow's own command, run on the server: ssh [email protected] "bash -l -c 'cd ~/aistack && git fetch origin main && git reset --hard origin/main && /root/.local/share/pnpm/pnpm i && /root/.local/share/pnpm/pnpm convex deploy'". Table rows survive; the indexes come back with the deploy.
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.
- 3d ago First seen · 348 lines · 4,908 tokens per session scan A 47c87de1752a
aistack AGENTS.md is an instructions file published in the GitHub repository alp82/aistack (47 stars, last pushed 3d ago), licensed MIT. It adds 4,908 tokens to every session, about $0.0245 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.
Other instructions, from other repositories
aider-desk AGENTS.md
AGENTS.md instructions for hotovo/aider-desk, covering agents.md, common commands, development, type checking and linting and formatting.
cctop AGENTS.md
Instructions for st0012/cctop, covering agents.md - development guide for cctop, scope and sources, required development rules, driver workflow and github and pr rules.
toh-framework CLAUDE.md
Instructions for wasintoh/toh-framework, covering claude.md — toh framework (repo development guide), what this is, everyday commands, verification protocol and single source, transformed per ide.
examples CLAUDE.md
Claude Code instructions for rossoctl/examples, covering claude.md - agent examples, repository structure, key commands, code style and dco sign-off (mandatory).
agent-rules child-process.instructions.md
Instructions for lirantal/agent-rules, covering system processes secure coding guidelines, your mission and spawning system processes.
claude-code-templates CLAUDE.md
Claude Code instructions for Justdvp/claude-code-templates, covering claude.md, project overview, development commands, package management and application commands.