inference-x AGENTS.md

A set of instructions for agents working on InferenceX, a project that uses OpenSpec to plan and track non-trivial changes before implementation. It also defines mandatory tooling paths intended to keep work efficient.

In plain words
What is it for?
Use it when planning, implementing, validating, or archiving changes in InferenceX. It requires the relevant OpenSpec artifacts and project documents to be read in a defined order.
Why use it?
It prevents undocumented implementation work and ensures that architecture, decisions, specifications, and review requirements are checked first.

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/coeusyk/inference-x/agents-md
Clone the repo
git clone --depth 1 https://github.com/coeusyk/inference-x

Made for: Codex, OpenCode.

Per session 2,642 This file is loaded in full into every session.
When invoked 2,642 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 $0.02642 $0.02642
Opus 5 $0.01321 $0.01321
Sonnet 5 $0.00528 $0.00528
Haiku 4.5 $0.00264 $0.00264

Measured 2d ago against content hash 8df1e20c15c4, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

inference-x 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 · 239 lines

How it starts

The opening of the file, as written. The whole thing — 239 lines — stays where its author put it; the contents beside it link to each section on GitHub.

OpenSpec Agent Instructions for InferenceX

This repository uses OpenSpec for spec-driven development.

Working agreement

  • Do not implement non-trivial work without an active change under openspec/changes/.
  • Before making changes, read:
    • docs/ARCHITECTURE.md
    • docs/PHASES.md
    • docs/DECISIONS.md
    • docs/REVIEW_MODE.md
    • docs/BUILD_MODE.md
    • relevant files in openspec/specs/
    • active change artifacts in openspec/changes/<change-id>/
  • For each change, work in this order:
    1. proposal.md
    2. spec deltas under specs/
    3. design.md
    4. tasks.md
    5. implementation
    6. validation
    7. archive

Agent tooling (mandatory — token efficiency)

Token efficiency is a standing requirement for every session in this repo, not a per-task preference. Prefer the smallest useful context: filter command output, sandbox analysis so only answers enter the conversation, and look up symbols through indexes instead of dumping files.

Two command paths only — no third

Every shell command must run through one of:

Path When Why
context-mode (ctx_execute, ctx_batch_execute, ctx_execute_file, …) Default for read-only work: gates (pytest/ruff/mypy), openspec validate, greps, inspections, scripts that only need stdout Sandboxes execution so only stdout/stderr you print enter the model context (context-mode targets ~98% reduction on tool output). Large results stay indexed; follow up with ctx_search.
rtk proxy <cmd> Host-state mutations that must persist: git commits/branches/merges/pushes, openspec archive, moves/renames outside Write/Edit, long-running processes whose exit code on the real host matters ctx_execute runs in a subprocess and discards filesystem writes. Mutations that must land in the real worktree leave context-mode via rtk proxy.

A bare Bash/Shell call, or bare rtk <cmd> without proxy when you meant a persisting mutation, is wrong for this repo.

Read the full file on GitHub · 239 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 · 239 lines · 2,642 tokens per session scan A 8df1e20c15c4

Subscribe to this mod's changes

inference-x AGENTS.md is an instructions file published in the GitHub repository coeusyk/inference-x (2 stars, last pushed 22d ago), licensed MIT. It adds 2,642 tokens to every session, about $0.0132 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.