taskyon AGENTS.md

Project instructions for working on Xyntopia and Taskyon, including rules for reading policies, checking current code and documentation, and handling external proposals.

In plain words
What is it for?
Use them when modifying agent instructions, proposals, policies, or other Taskyon-related project files.
Why use it?
They help an agent make changes using the project's governing policies and current implementation evidence instead of assumptions.

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/xyntopia/taskyon/agents-md
Clone the repo
git clone --depth 1 https://github.com/Xyntopia/taskyon

Made for: Codex, OpenCode.

Per session 2,352 This file is loaded in full into every session.
When invoked 2,352 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 $0.02352 $0.02352
Opus 5 $0.01176 $0.01176
Sonnet 5 $0.00470 $0.00470
Haiku 4.5 $0.00235 $0.00235

Measured 2d ago against content hash 54fe25b260c5, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

taskyon 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.

AGENTS.md · 189 lines

How it starts

The opening of the file, as written. The whole thing — 189 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Instructions

Read the canonical Taskyon Core Policies before every change. They are the public project philosophy and apply to agents and human contributors alike.

The operational rules in this file apply to every agent change. The routing table at the end points to additional stable policies for specialized work.

Do not use this file as evidence of current commands, package layout, versions, feature support, or implementation status. Inspect the current owning code, configuration, package README, or active developer documentation for those changing facts.

When project principles compete, use the order in the core policy document. Correctness and the user's requested scope take precedence over implementation preferences. State any material tradeoff.

External Proposal Ownership

Taskyon is an independent repository. Architectural proposals are maintained outside this source repository, and this project does not assume a particular parent project, deployment repository, or proposal path. Before creating or editing a proposal, ask the maintainer which external repository and path should own it. Keep Taskyon policies and documentation focused on implementation invariants and current behavior; do not add deployment-specific proposal paths or links here.

Protocol Discipline

  • Treat every protocol command as a capability that could be exposed across a future P2P boundary.
  • Use typed protocols at real runtime, service, storage, and ownership boundaries even when the current provider is local or in-process. Consumers must not depend on whether the implementation runs locally, in a worker, on another process, or on a peer.
  • Keep client behavior and protocol semantics location-transparent. Select and connect an adapter at the composition boundary instead of branching throughout consumers on local versus remote placement.
  • Keep protocols focused on intentional remote capabilities and canonical data exchange. Do not add graph traversal, filtering, ranking, aggregation, or other convenience queries merely to simplify a local consumer.
  • Before extending a protocol, prove that the behavior cannot be derived cleanly from locally available canonical records using a cache or rebuildable index.
  • When local data is incomplete, prefer synchronizing or fetching the smallest canonical records needed and derive views locally instead of adding a specialized remote query.
  • Expose derived computation through a protocol only when performing that computation remotely is itself the intended, capability-scoped service, with explicit trust and authorization semantics.
  • Do not turn ordinary calls within one cohesive implementation into protocol traffic. Keep local implementation details as direct typed functions and introduce a protocol where it removes a genuine ownership, lifecycle, runtime, trust, or replaceability coupling.

Read the full file on GitHub · 189 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. 2d ago First seen · 189 lines · 2,352 tokens per session scan A 54fe25b260c5

Subscribe to this mod's changes

taskyon AGENTS.md is an instructions file published in the GitHub repository Xyntopia/taskyon (54 stars, last pushed 13d ago), licensed MIT. It adds 2,352 tokens to every session, about $0.0118 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.