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/cagataycali/devduck/agents-mdgit clone --depth 1 https://github.com/cagataycali/devduckWhat 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 | $0.01592 | $0.01592 |
| Opus 5 | $0.00796 | $0.00796 |
| Sonnet 5 | $0.00318 | $0.00318 |
| Haiku 4.5 | $0.00159 | $0.00159 |
Grade A, and why
devduck 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 — 136 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — DevDuck
Project: DevDuck
- What: Self-modifying AI agent that hot-reloads its own code
- Language: Python 3.10–3.13
- Framework: Strands Agents SDK
- Entry point:
devduck/__init__.py(single-file core) - Package:
pip install devduck/pipx install devduck - License: Apache 2.0
Architecture
DevDuck is a single DevDuck class in __init__.py that auto-initializes on import.
The module itself is callable: import devduck; devduck("query").
devduck/
├── __init__.py # Core agent: DevDuck class, REPL, CLI, session recording, ambient mode
├── tui.py # Multi-conversation Textual TUI (concurrent panels, streaming markdown)
├── landing.py # Rich landing screen for REPL mode
├── callback_handler.py # Streaming callback handler for CLI
├── asciinema_callback_handler.py # Callback handler that records .cast files
├── agentcore_handler.py # HTTP handler for Bedrock AgentCore deployment
├── tools/ # 60+ built-in tools
│ ├── system_prompt.py # Self-improvement via prompt management
│ ├── manage_tools.py # Runtime tool add/remove/create/fetch
│ ├── manage_messages.py # Conversation history management
│ ├── websocket.py # WebSocket server (per-message agent)
│ ├── zenoh_peer.py # P2P auto-discovery networking
│ ├── agentcore_proxy.py # Unified mesh relay (Zenoh + AgentCore + browser)
│ ├── unified_mesh.py # Ring context shared memory
│ ├── mesh_registry.py # File-based agent discovery with TTL
│ ├── tasks.py # Background parallel agent tasks
│ ├── scheduler.py # Cron and one-time job scheduling
│ ├── telegram.py # Telegram bot integration
│ ├── slack.py # Slack integration
│ ├── whatsapp.py # WhatsApp via wacli
│ ├── tui.py # Agent-side TUI content push
│ ├── speech_to_speech.py # Real-time voice (Nova Sonic, OpenAI, Gemini)
│ ├── lsp.py # Language Server Protocol diagnostics
│ ├── use_mac.py # Unified macOS control
│ ├── apple_notes.py # Apple Notes management
│ ├── use_spotify.py # Spotify control
│ └── ... # tcp, ipc, mcp_server, scraper, rl, etc.
└── tools/ (hot-reload) # ./tools/*.py auto-loaded at runtime
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 · 136 lines · 1,592 tokens per session scan A e4046879a71a
devduck AGENTS.md is an instructions file published in the GitHub repository cagataycali/devduck (42 stars, last pushed 3mo ago), licensed Apache-2.0. It adds 1,592 tokens to every session, about $0.0080 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
containerized-strands-agents AGENTS.md
Instructions for mkmeral/containerized-strands-agents, covering project overview, repository structure, build & development commands, install in development mode and install with web ui support.
mcp-supersubagents CLAUDE.md
Claude Code instructions for yigitkonur/mcp-supersubagents, covering claude.md, what this is, build & run, environment variables and pat tokens (checked in priority order).
mcp-supersubagents AGENTS.md
AGENTS.md instructions for yigitkonur/mcp-supersubagents, covering agent guidelines, architecture, non-negotiable rules, coding standards and workflow.
mcp-supersubagents copilot-instructions.md
Copilot instructions for yigitkonur/mcp-supersubagents, covering mcp super-subagents — code review standards, stdio transport integrity, security, error handling tiers and state mutation safety.
mcp-supersubagents mcp-tools.instructions.md
Instructions for yigitkonur/mcp-supersubagents, covering mcp tool handler guidelines, input validation, spawn handler pattern, error handling and response format.
mcp-supersubagents services.instructions.md
Instructions for yigitkonur/mcp-supersubagents, covering service layer guidelines, singleton pattern, task state machine, in-place mutation — load-bearing and async race prevention.