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/lathe-cli/kitup/agents-mdgit clone --depth 1 https://github.com/lathe-cli/kitupWhat 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 | $0.01531 | $0.01531 |
| Opus 5 | $0.00766 | $0.00766 |
| Sonnet 5 | $0.00306 | $0.00306 |
| Haiku 4.5 | $0.00153 | $0.00153 |
Grade A, and why
kitup 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.
How it starts
The opening of the file, as written. The whole thing — 155 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Project Boundary
kitup is a producer-side SDK for CLI authors who ship bundled Agent Skills with their tools.
The v0.1 product is deliberately small:
- validate a local skill directory
- validate an embedded bundled skill directory tree
- resolve a public GitHub skill bundle directory tree
- detect local agent hosts
- resolve safe CLI install selection before writes
- run the safe CLI install workflow for bundled skills
- resolve user and project skill directories
- copy bundled skill files
- write
.kitup.jsonownership metadata - update or uninstall only kitup-owned installs
- refuse unsafe overwrite conflicts
- return structured reports
- enforce parity through shared golden cases
Do not turn kitup into a marketplace, registry client, package manager, MCP server, GUI, agent runtime integration layer, or arbitrary skill installer. Only first-party public GitHub bundle resolution is inside v0.1 behavior. Other remote installs, private GitHub auth, marketplace search, plugin systems, AGENTS.md mutation, script execution, and symlink-first workflows are outside v0.1 unless this boundary and golden cases are changed first.
Product Promise
One bundled skill, one SDK call, many agent hosts.
The embedding CLI owns the skill and command shell. kitup owns the boring installer layer: host path lookup, detection, selection prompts, summary confirmation, validation, copy/update/uninstall semantics, metadata, conflicts, and reports.
Keep the developer experience boring. A CLI author should only need to provide:
appIdskillBundlescopeagents
Everything else must be derived from shared data and deterministic installer behavior.
Source Of Truth
Use the live repository state, not memory or aspiration.
spec/hosts.jsonis the host adapter database.spec/*.schema.jsondefines accepted shared data shape.testdata/cases/*.jsondefines cross-language behavior.- Generated host constants in
ts/,go/,rust/, andpython/come fromspec/hosts.json; never edit them by hand. scripts/sync-hosts.mjsrefreshes and checks generated host constants.scripts/check.mjsis the current parity gate.docs/host-adapter-contract.mdexplains host adapter semantics.README.mddocuments the public product boundary; code, schemas, and golden cases are the enforcement layer.
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.
- yesterday First seen · 155 lines · 1,531 tokens per session scan A 0b805e634ea9
kitup AGENTS.md is an instructions file published in the GitHub repository lathe-cli/kitup (6 stars, last pushed 4d ago), licensed Apache-2.0. It adds 1,531 tokens to every session, about $0.0077 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
openapi CLAUDE.md
Instructions for longbridge/openapi, covering agent guidelines, after modifying rust code, after modifying the node.js sdk (nodejs/), after modifying the python sdk api (python/) and after modifying the c sdk (c/).
openapi AGENTS.md
Instructions for longbridge/openapi, covering agent guidelines, after modifying rust code, after modifying the node.js sdk (nodejs/), after modifying the python sdk api (python/) and after modifying the c sdk (c/).
Better-Fullstack AGENTS.md
Instructions for Marve10s/Better-Fullstack, covering guidelines, web ui, workflow and bun.
rust-mcp-sdk CLAUDE.md
Instructions for paiml/rust-mcp-sdk, covering pmcp sdk development standards, toyota way quality system - zero tolerance for defects, quality gate enforcement, pre-commit quality gates (mandatory) and ci quality gates (pr-blocking, added phase 75 wave 5).
Better-Fullstack CLAUDE.md
Instructions for Marve10s/Better-Fullstack, covering claude.md, code style, general preferences and picking the right models for workflows and subagents.
guff AGENTS.md
AGENTS.md instructions for dakimura/guff: When working in the guff codebase, follow docs/DEVELOPMENT.md and the compat isolate gate in .cursor/rules.