minutes AGENTS.md

Project instructions for AI assistants working on the silverstein/minutes repository. They explain its local issue tracker, GitHub Discussions, and how to keep generated outputs current.

In plain words
What is it for?
Use them when working on issues, answering usage questions, updating user-facing help, or regenerating project files.
Why use it?
They give an assistant the project rules and support conventions needed to make changes consistently.

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/silverstein/minutes/agents-md
Clone the repo
git clone --depth 1 https://github.com/silverstein/minutes

Made for: Codex, OpenCode.

Per session 3,234 This file is loaded in full into every session.
When invoked 3,234 The same file — it is already loaded in full.
Security scan C 2 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 $0.03234 $0.03234
Opus 5 $0.01617 $0.01617
Sonnet 5 $0.00647 $0.00647
Haiku 4.5 $0.00323 $0.00323

Measured 2d ago against content hash 8b2c3431d2e9, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade C, and why

minutes AGENTS.md scanned grade C 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.

Recursive force deletehighDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

rm -rf directory # NOT: rm -r directory

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

PUBLISHED=$(curl -s https://crates.io/api/v1/crates/whisper-guard | jq -r '.crate.max_stable_version')
AGENTS.md · 272 lines

How it starts

The opening of the file, as written. The whole thing — 272 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Instructions

This project uses bd (beads) for issue tracking. Run bd onboard to get started. In this repo, beads is local-only: use it for structured issue tracking on your machine, but do not expect a shared Dolt remote.

Quick Reference

bd ready              # Find available work
bd show <id>          # View issue details
bd update <id> --claim  # Claim work atomically
bd close <id>         # Complete work

GitHub Discussions

This repo has GitHub Discussions enabled (silverstein/minutes). Issues are for bugs and feature requests. Discussions are for usage questions, setup help, and community show-and-tell.

Agent guidelines:

  • When triaging an issue that's really a "how do I...?" question, suggest converting it to a Discussion rather than closing it
  • When a user's bug report turns out to be a config/setup issue, answer it and note that Discussions is the better venue for follow-ups
  • After shipping a feature or fix, check if any open Q&A discussions are resolved by the change — post a reply pointing to the release
  • When writing user-facing error messages or help text, link to Discussions (not Issues) for support: https://github.com/silverstein/minutes/discussions
  • Don't file Discussions as work items — they're community conversations, not tracked tasks

Portable Agent Skills (.agents/skills/minutes/, .opencode/skills/)

This repo maintains skill outputs in three locations:

  • .claude/plugins/minutes/ — Claude Code plugin (uses ${CLAUDE_PLUGIN_ROOT})
  • .agents/skills/minutes/ — Agent-agnostic mirror for Codex, Gemini, Pi, and other agents (uses $MINUTES_SKILLS_ROOT)
  • .opencode/skills/ — OpenCode-native mirror (one-level discovery path + matching .opencode/commands/)

What lives where:

  • SKILL.md files are mirrored 1:1. Content is identical except for path variables and platform-specific references (e.g., "open in desktop app" in the plugin version becomes a CLI command in the agents version).
  • _runtime/hooks/lib/ contains minutes-learn.mjs and minutes-learn-cli.mjs — the behavioral learning system. These must stay byte-identical across .agents/skills/minutes/_runtime/hooks/lib/ and .opencode/skills/_runtime/hooks/lib/.
  • Bundled scripts (for example scripts/tag_apply.py) are mirrored into both portable trees.
  • .opencode/commands/*.md provides native /minutes-* slash commands for OpenCode and is generated from the same canonical skill sources.

Read the full file on GitHub · 272 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 · 272 lines · 3,234 tokens per session scan C 8b2c3431d2e9

Subscribe to this mod's changes

minutes AGENTS.md is an instructions file published in the GitHub repository silverstein/minutes (1,458 stars, last pushed today), licensed MIT. It adds 3,234 tokens to every session, about $0.0162 per session on Opus 5. A static security scan graded it C with 2 findings (recursive force delete, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.