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/numikel/kprun/claude-mdgit clone --depth 1 https://github.com/numikel/kprunWrote 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/numikel/kprun/claude-md)<a href="https://agentmods.dev/instructions/numikel/kprun/claude-md"><img src="https://agentmods.dev/badge/instructions/numikel/kprun/claude-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 | $0.02080 | $0.02080 |
| Opus 5 | $0.01040 | $0.01040 |
| Sonnet 5 | $0.00416 | $0.00416 |
| Haiku 4.5 | $0.00208 | $0.00208 |
Grade A, and why
kprun CLAUDE.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 — 135 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
Superpowers workflow
The user invokes all superpowers workflow skills (brainstorming → writing-plans → executing-plans etc.) manually. After finishing one stage (e.g. saving an approved design spec), STOP — do not ask whether to proceed and do not auto-invoke the next skill in the chain.
Model tier mapping
When any Superpowers skill (especially subagent-driven-development,
"Model Selection" section) refers to a "cheap/mechanical model", "standard
model", or "most capable model" when dispatching a subagent, do NOT leave
this underspecified or silently inherit the session's default model.
Substitute the values below directly into the model: field (and
effort:, if that field exists in the dispatch template).
| Skill tier | Model | Effort |
|---|---|---|
| cheap / mechanical | claude-haiku-4-5 | — |
| standard / execution | claude-sonnet-5 | medium |
| most capable / review | claude-opus-4-8 | xhigh |
Escalation on BLOCKED or repeated failure
claude-haiku-4-5 → claude-sonnet-5 (effort: high) → claude-opus-4-8 (effort: xhigh). Never redispatch the same task on the same model without changing approach — escalate the tier instead of repeating the attempt.
Exceptions to the default step mapping (always "most capable",
regardless of which step they appear in)
- systematic-debugging (root-cause on a hard bug)
- final whole-branch code reviewer (superpowers:requesting-code-review)
- brainstorming and writing-plans
Exceptions that are always "cheap", regardless of the general step mapping
- using-git-worktrees
- finishing-a-development-branch
- /prepare-release — deterministic checklist; escalate to claude-sonnet-5 (effort: medium) ONLY if the skill's §6 self-check (cargo fmt / cargo test / version consistency) fails on the first pass
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 · 135 lines · 2,080 tokens per session scan A 428a5b8c8e13
kprun CLAUDE.md is an instructions file published in the GitHub repository numikel/kprun (9 stars, last pushed 10d ago), licensed MIT. It adds 2,080 tokens to every session, about $0.0104 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
kdbx AGENTS.md
Instructions for yarrasys/kdbx, covering agents.md, what this is, current status — read this first, golden rules and repository map.
kdbx CLAUDE.md
Instructions for yarrasys/kdbx, a project described as: Per-project secrets in local KeePassXC vaults, injected into commands without ever printing them. Agent-safe: PreToolUse guard hook + read-only MCP server. Single Go binary.
lade AGENTS.md
AGENTS.md instructions for zifeo/lade, covering agents.md, build & test, house rules, project layout and maintainer note.
agenv CLAUDE.md
Instructions for combinatrix-ai/agenv, covering claude.md, commands, do not modify, architecture and conventions.
env.sync.local AGENTS.md
AGENTS.md instructions for championswimmer/env.sync.local, covering agents.md - project guide for llm coding agents, project overview, architecture, core philosophy and three security modes (v3.0+).
env.sync.local CLAUDE.md
Claude Code instructions for championswimmer/env.sync.local, a project described as: sync your env secrets securely peer-to-peer on local networks.