claude-code-telegram-kit AGENTS.md

Repository-specific instructions stored in an AGENTS.md file for the claude-code-telegram-kit project. They define design rules, security boundaries, and changes to avoid.

In plain words
What is it for?
Use them when modifying that repository to check its design invariants, prohibited changes, and required security practices.
Why use it?
They help coding agents make changes that preserve the project's deterministic Telegram delivery, session control, and filesystem safety behavior.

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/project-tharsis/claude-code-telegram-kit/agents-md
Clone the repo
git clone --depth 1 https://github.com/project-tharsis/claude-code-telegram-kit

Made for: Codex, OpenCode.

Per session 510 This file is loaded in full into every session.
When invoked 510 The same file — it is already loaded in full.
Security scan A 0 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 $0.00510 $0.00510
Opus 5 $0.00255 $0.00255
Sonnet 5 $0.00102 $0.00102
Haiku 4.5 $0.00051 $0.00051

Measured yesterday against content hash 863c86914b19, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

claude-code-telegram-kit 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 yesterday.

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.

AGENTS.md · 41 lines

How it starts

The opening of the file, as written. The whole thing — 41 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Working on this repository

This kit exists so that a model chooses nothing about Telegram delivery or session control. Hooks and deterministic code decide; Claude only returns canonical Markdown. Most changes that look like improvements erode that property. Read docs/design-invariants.md before proposing one.

Do not

  • Add a model-facing tool. Transport selection, quote target, fallback, reactions, artifact discovery, and disclosure redaction are deterministic and stay that way. Any new model-facing surface requires an explicit design-invariant change and security review before implementation.
  • Retry an uncertain outcome, or convert one into a failure. A timeout, 429, 5xx, or unknown response leaves state untouched and leaves 👀 in place. Only proven local or permanent rejection may finalize 👎.
  • Weaken a filesystem check. Preserve directory-FD anchoring, owner, mode, inode, and link-count verification wherever the current boundary uses them. Do not replace those checks with a bare path, realpath, or a single stat.
  • Give the unprivileged side privileged authority. It never executes sudo, systemd-run, helper or config paths, unit names, or arbitrary argv. Privileged work crosses the peer-UID-checked broker socket and nothing else.
  • Commit real deployment identity. Paths, chat IDs, bot usernames, and tokens stay out of the tree. scripts/check_public_tree.py is a backstop, not a substitute for care.

Do

  • Run bun run check before claiming anything works. It covers tests, typecheck, Python compile and unit tests, the privacy scan, and version agreement.
  • Version each wire contract explicitly. A Broker Protocol request/response change must update the TypeScript client and root broker together; a Session Control Protocol or helper CLI contract change must update the broker capability expectation and helper together.
  • Keep the two envelope parsers in sync through packages/shared/fixtures/telegram-envelope-cases.json. The command guard must never allow a direct control envelope that the strict TypeScript parser accepts to fall through to the model.
  • Change release versions in one operation. Four package.json files, bun.lock, and both src/server.ts identities must match; scripts/check_versions.py enforces the source identities.

Read the full file on GitHub · 41 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. yesterday First seen · 41 lines · 510 tokens per session scan A 863c86914b19

Subscribe to this mod's changes

claude-code-telegram-kit AGENTS.md is an instructions file published in the GitHub repository project-tharsis/claude-code-telegram-kit (0 stars, last pushed 5d ago), licensed Apache-2.0. It adds 510 tokens to every session, about $0.0026 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.