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/graniet/kheish/agents-mdgit clone --depth 1 https://github.com/graniet/kheishWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/graniet/kheish/agents-md)<a href="https://agentmods.dev/instructions/graniet/kheish/agents-md"><img src="https://agentmods.dev/badge/instructions/graniet/kheish/agents-md.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.04148 | $0.04148 |
| Opus 5 | $0.02074 | $0.02074 |
| Sonnet 5 | $0.00830 | $0.00830 |
| Haiku 4.5 | $0.00415 | $0.00415 |
Grade A, and why
kheish 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 — 617 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Kheish Operator Guide
This file is for future agent instances working inside kheish/.
Read this first when you need to change code, run the daemon, validate behavior on real providers, or debug a broken session.
This is an operational guide, not a full architecture spec.
Working Rules
- Treat the daemon as the source of truth.
- Prefer the real CLI and HTTP API over direct inspection of JSON state files.
- Use fresh
--state-rootvalues for real validation. - Use fresh sessions for E2E tests.
- Validate on the real daemon, not only with unit tests.
- Keep examples and paths portable. This repository may be cloned anywhere.
- When behavior differs across providers, keep the abstraction provider-neutral unless the provider wire format truly requires divergence.
- Do not infer product behavior from stale sessions created by an older daemon build.
Repository Basics
Repository root:
- the directory that contains this
AGENTS.md
Important files:
.envCargo.tomlREADME.mdAGENTS.md
Most active crates:
crates/kheish-daemon: daemon HTTP API, CLI integration, run/session/task control planecrates/kheish-runtime: provider drivers, prompt building, permissions, runtime assemblycrates/kheish-core: engine loop, compaction, hook integrationcrates/kheish-coding-tools: built-in tools such asread_file,write_file,bash,web_fetch,web_searchcrates/kheish-mcp: MCP loading, config import, tool/resource adaptationcrates/kheish-agent: agent orchestration, snapshots, sidechains, mailboxcrates/kheish-types: shared data contracts
Important old state roots that may still exist in this repo:
.kheish-daemon.kheish-daemon2.kheish-daemon-fix.kheish-daemon-fix2
Do not assume the daemon you are talking to uses the state root you expect. Always be explicit.
Environment
Load provider keys before live testing:
cd "$(git rev-parse --show-toplevel)"
set -a
source .env
set +a
Supported providers in normal operation:
anthropicgoogleopenaiopenrouterxai
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 · 617 lines · 4,148 tokens per session scan A 625f39a4936d
kheish AGENTS.md is an instructions file published in the GitHub repository graniet/kheish (227 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 4,148 tokens to every session, about $0.0207 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
eve AGENTS.md
Instructions for vercel/eve, covering agents.md, about eve, repository layout, git workflow and commands.
AgentEval memory-benchmarks.instructions.md
Guidelines for working with AgentEval Memory module — benchmarks, reporting, HTML reports, and LongMemEval integration.
AgentEval copilot-instructions.md
Instructions for AgentEvalHQ/AgentEval, covering agenteval - ai coding agent instructions, architecture overview, environment setup, optional: secondary models for comparison and build & test commands.
AgentEval maf-updates.instructions.md
Instructions for detecting and adapting to Microsoft Agent Framework (MAF) breaking changes.
AgentEval maf-upgrade-preparation.instructions.md
Instructions for analyzing a new MAF version and producing an upgrade plan BEFORE updating the NuGet package.
AgentEval AGENTS.md
Instructions for AgentEvalHQ/AgentEval, covering agenteval development instructions, project overview, quick reference, build & test and key directories.