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/littlebearapps/untether/agents-mdgit clone --depth 1 https://github.com/littlebearapps/untetherWhat 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 | $0.00709 | $0.00709 |
| Opus 5 | $0.00354 | $0.00354 |
| Sonnet 5 | $0.00142 | $0.00142 |
| Haiku 4.5 | $0.00071 | $0.00071 |
Grade A, and why
untether 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 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.
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 — 78 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Untether — Agent Instructions
Telegram bridge for AI coding agents. Control Claude Code, Codex, OpenCode, Pi, Gemini CLI, and Amp from your phone or any device — agents run on your machine in the background while you're away from the terminal. Features interactive permissions, voice input, cost tracking, and live progress streaming.
Architecture
Telegram <-> TelegramPresenter <-> RunnerBridge <-> Runner (claude/codex/opencode/pi/gemini/amp)
|
ProgressTracker
- Runners (
src/untether/runners/) — engine-specific subprocess managers - RunnerBridge (
src/untether/runner_bridge.py) — connects runners to Telegram presenter - TelegramPresenter (
src/untether/telegram/bridge.py) — renders progress, inline keyboards, answers - Commands (
src/untether/telegram/commands/) — command/callback handlers - Schemas (
src/untether/schemas/) — msgspec structs for JSONL parsing
Key conventions
- Python 3.12+, anyio for async, msgspec for JSONL, structlog for logging
- Ruff for linting (
uv run ruff check src/), pytest with 80% coverage threshold - Australian English in user-facing text (realise, colour, behaviour, licence)
- Conventional commits:
feat:,fix:,docs:,refactor:,test: - Feature branches:
feature/*,fix/*,docs/*
Runner 3-event contract
Every run MUST emit exactly:
StartedEvent— once, when session ID is knownActionEvent(s)— zero or moreCompletedEvent— exactly once, always final
Use EventFactory for event construction. Never construct event dataclasses directly.
Telegram transport rules
- ALL writes go through
TelegramOutbox(never call Bot API directly) - Callback data max 64 bytes, format:
prefix:action:id - Call
answerCallbackQuerypromptly to clear button spinners - Message limit 4096 chars; Untether splits long responses across multiple messages by default
Testing
uv run pytest # all tests
uv run pytest tests/test_*.py -x # specific file
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 · 78 lines · 709 tokens per session scan A f75af8c61da7
untether AGENTS.md is an instructions file published in the GitHub repository littlebearapps/untether (66 stars, last pushed 3d ago), licensed MIT. It adds 709 tokens to every session, about $0.0035 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
solo AGENTS.md
Instructions for solo-agent/solo, covering project testing rules and project service lifecycle rules.
superset AGENTS.md
AGENTS.md instructions for superset-sh/superset, covering superset monorepo, project structure, exception: shadcn/ui components, database and releases.
agent-tools AGENTS.md
Instructions for kairyou/agent-tools, covering agent instructions, build and test, layout, settled decisions and docs.
superset CLAUDE.md
Claude Code instructions for superset-sh/superset, a project described as: Superset is an agentic IDE to orchestrate 100+ coding agents in parallel. Run any agent with your own subscription.
awesome-copilot-id AGENTS.md
Instructions for GulajavaMinistudio/awesome-copilot-id, covering communication, explanation and documentation, markdown formatting, user communication style and workflow & methodology.
sessionview AGENTS.md
Instructions for tyql688/sessionview, covering sessionview, commands, app / frontend, gates must pass under both feature sets: default (gui) and and --no-default-features --features headless.