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/yashau/dexgram/agents-mdgit clone --depth 1 https://github.com/yashau/dexgramWhat 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 | $0.01829 | $0.01829 |
| Opus 5 | $0.00915 | $0.00915 |
| Sonnet 5 | $0.00366 | $0.00366 |
| Haiku 4.5 | $0.00183 | $0.00183 |
Grade A, and why
dexgram 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 — 202 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Guidance for agents working on Dexgram.
Project Overview
Dexgram is a Windows-only Telegram bridge for Codex Desktop. It runs as a Go binary, listens through a Telegram bot with threaded topics enabled, and maps Telegram topics to Codex app-server threads.
Dexgram owns one shared Codex app-server --listen stdio:// process per
Dexgram runtime. Topic sessions are routing records over that shared client, not
one app-server per Telegram topic. Events and approval/user-input requests must
be routed by Codex threadId/turnId to the mapped Telegram topic. The shared
app-server is tracked with a small PID marker and guarded by a Windows Job
Object so stale children are cleaned up after crashes or restarts.
Side chats are Telegram topics created with /side or /btw. They are backed
by native Codex thread/fork, keep the same project/cwd as the parent, and use
a ↳N topic-name prefix so they do not get confused with the main topic.
Existing Codex sessions can be attached to Telegram topics through
/sessions [query]. The browser lists projects first, then paginated threads
inside a selected project. Attaching a session maps that topic to the existing
Codex thread and syncs up to the most recent 25 rendered Telegram history
messages.
Fresh unbound topics must not silently become projectless chats on the first prompt. Dexgram should first ask whether to resume an existing session, start a new chat, or set a project first.
The main command lives in cmd/dexgram. Shared packages live under internal/:
internal/codex: Codex app-server JSON-RPC client.internal/codexprojects: Codex Desktop project discovery.internal/codexstate: projectless workspace helpers.internal/config: TOML config loading/defaults.internal/state: SQLite-backed mapping, sync, settings, and pairing-code state.
Keep changes small and aligned with the current layout. Prefer local helpers and existing package boundaries over adding new framework-like abstractions.
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 · 202 lines · 1,829 tokens per session scan A 89d1c8792499
dexgram AGENTS.md is an instructions file published in the GitHub repository yashau/dexgram (7 stars, last pushed 2mo ago), licensed MIT. It adds 1,829 tokens to every session, about $0.0091 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 2026-08-31.
Other instructions, from other repositories
LangBot AGENTS.md
Instructions for langbot-app/LangBot, covering agents.md, quick facts, essential commands, where to look and cross-repo sdk work.
acodex AGENTS.md
Instructions for maksimzayats/acodex, a project described as: Local MCP automation bridge for the Codex desktop app, with CLI, HTTP server, and live codexapp. tool access.
arcvellum AGENTS.md
Instructions for o-1717986918/arcvellum, covering agents, product boundary, development entry, worker constitution and verification.
atifact AGENTS.md
Instructions for waldekmastykarz/atifact, covering agents and atif spec version.
LangBot CLAUDE.md
Instructions for langbot-app/LangBot, a project described as: Production-grade platform for building agentic IM bots - 生产级多平台智能机器人开发平台/ Agent、知识库编排、插件系统 / Bots for Discord / Slack / LINE / Telegram / WeChat(企业微信, 企微智能机器人, 公众号) / 飞书 / 钉钉 / QQ / Matrix e.g. Integrated with ChatGPT(GPT), DeepSeek, Dify, n8n, Langflow…
codexcw AGENTS.md
AGENTS.md instructions for c3-oss/codexcw, covering agents, project shape, build, test, develop, coding style and commits and prs.