heatherbot AGENTS.md

A setup guide for HeatherBot, a locally run Telegram companion that uses a real Telegram account and a local language model.

In plain words
What is it for?
It helps agents install and start the text-based bot, then understand optional services for images, voice, memory, and Discord.
Why use it?
It explains the required software, credentials, configuration, and startup order, so an agent can prepare a new machine without missing prerequisites.

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

Made for: Codex, OpenCode.

Per session 4,021 This file is loaded in full into every session.
When invoked 4,021 The same file — it is already loaded in full.
Security scan A 1 finding. 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 $0.04021 $0.04021
Opus 5 $0.02011 $0.02011
Sonnet 5 $0.00804 $0.00804
Haiku 4.5 $0.00402 $0.00402

Measured 3d ago against content hash 3b3cc4ae68c2, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

heatherbot AGENTS.md scanned grade A with 1 finding 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.

Makes network callslowCapability

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

curl http://localhost:1234/v1/models
AGENTS.md · 347 lines

How it starts

The opening of the file, as written. The whole thing — 347 lines — stays where its author put it; the contents beside it link to each section on GitHub.

AGENTS.md — Bootstrap Guide for Coding Agents

This file is written for an autonomous coding agent (Claude Code, Codex, etc.) tasked with getting HeatherBot running on a fresh machine. It is a literal, ordered runbook. A human can follow it too. Read the whole file before you start executing — several steps have prerequisites that are cheaper to satisfy up front.


0. TL;DR for an agent

HeatherBot is a fully local, persona-driven AI companion that runs as a Telethon MTProto userbot (it logs in as a real Telegram account, not a Bot API bot). It talks to a local LLM over an OpenAI-compatible HTTP endpoint and has optional local services for image analysis, image generation, and voice.

To get a minimal text-only bot alive you need exactly four things:

  1. Python 3.11
  2. A running local LLM server on http://localhost:1234/v1 (llama.cpp's llama-server, LM Studio, or anything OpenAI-compatible)
  3. Telegram API credentials (API_ID + API_HASH) and a phone number to log in
  4. A persona YAML (ship persona_example.yaml as-is to start)

Everything else (Ollama vision, ComfyUI image gen, ElevenLabs/Coqui voice, semantic vector memory, the Discord bridge) is optional and fail-soft — the core bot runs without them.

Do not expect a one-command install. This is infrastructure, not an app.


1. System requirements

Requirement Minimum Notes
OS Windows 10/11, Linux, or macOS Original deployment is Windows 11. Paths in helper .ps1 scripts are Windows; the Python is cross-platform.
Python 3.11.x 3.12 likely works; 3.11 is tested.
RAM 16 GB 32 GB+ comfortable.
GPU 1× 8 GB VRAM for a small (7B–12B) model A 24B model wants ~24 GB (one RTX 3090) or quantization. Original runs dual RTX 3090.
Disk 10–40 GB Dominated by the GGUF model file, which you supply separately.
Network Outbound to Telegram All AI inference is local; no cloud LLM required.

Read the full file on GitHub · 347 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. 3d ago First seen · 347 lines · 4,021 tokens per session scan A 3b3cc4ae68c2

Subscribe to this mod's changes

heatherbot AGENTS.md is an instructions file published in the GitHub repository dvoraknc/heatherbot (20 stars, last pushed 2mo ago), licensed MIT. It adds 4,021 tokens to every session, about $0.0201 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.