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 skills/tenstorrent/tt-studio/cli-designnpx skills add tenstorrent/tt-studio --skill cli-designgit clone --depth 1 https://github.com/tenstorrent/tt-studioWrote 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/skills/tenstorrent/tt-studio/cli-design)<a href="https://agentmods.dev/skills/tenstorrent/tt-studio/cli-design"><img src="https://agentmods.dev/badge/skills/tenstorrent/tt-studio/cli-design.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 5.1 | $0.00199 | $0.03579 |
| Opus 5 | $0.00100 | $0.01790 |
| Sonnet 5 | $0.00040 | $0.00716 |
| Haiku 4.5 | $0.00020 | $0.00358 |
Grade B, and why
cli-design scanned grade B with 1 finding 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 6d 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.
Asks for rootmediumPrivilege escalation
A mod that escalates privileges can change anything on the machine, not only the project.
- **Link docs; don't offer to fix the user's machine.** No `sudo service docker start` How it starts
The opening of the file, as written. The whole thing — 263 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLI design
How to build a CLI that orchestrates messy tools (Docker/compose, package installs, downloads, background services) and still reads like something a person designed. The tools emit thousands of lines; the user needs about twelve. This is the whole design — command surface, run structure, output, failures, checks, testing — and nothing in it is specific to one repo.
Install into another repo
python3 install.py ../other-repo # skill + reference files
python3 install.py ../other-repo --module-dest mycli/console.py # …and wire the module in
python3 install.py --user # available in every repo on this machine
Then see the design run before adopting it:
python3 reference/demo.py # the output language, ~15s (--fail for failure paths, -v for verbose)
python3 reference/cli_skeleton.py --help # the command surface (then: no flags, --info, --stop)
| File | What it gives you |
|---|---|
reference/cli_skeleton.py |
a working CLI in this design: bootstrap → parse → early dispatch → phases → ready card |
reference/console.py |
the output layer (Rich only): theme, step(), phases, activity row, cards |
reference/demo.py |
runnable tour of every output element; doubles as a smoke test |
reference/patterns.md |
the four failure/progress patterns, with before/after and tests |
reference/test_cli_output.py |
copy into tests/: pure helpers, stream parsing, PTY + non-TTY render checks |
Reference implementation in the wild: TT-Studio's tt_setup/ (+ its house-rules doc
dev-docs/launcher-terminal-design.md).
1. The shape of the CLI
One entrypoint. python run.py (or mycli) does everything: setup, start, stop,
purge, logs, status, bug reports. Not a script per task, not a Makefile the user has
to read. One name to remember, discoverable from --help.
It bootstraps itself. First run creates the managed venv, installs deps, and re-execs. The user needs a python3, nothing else. That bootstrap code runs before your dependencies exist — stdlib only, and it must never import your pretty layer. Mimic the style by hand there (a plain stdlib spinner is ~15 lines).
What ships with it
6 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 6d ago First seen · 263 lines · 199 tokens per session scan B 98b7103debc3
cli-design is a skill published in the GitHub repository tenstorrent/tt-studio (49 stars, last pushed today), licensed Apache-2.0. It adds 199 tokens to every session and 3,579 once invoked, about $0.0010 per session on Opus 5. A static security scan graded it B with 1 finding (asks for root). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
compiler-orchestrator
Orchestrate the Rust compiler port end-to-end. Discovers the current frontier, fixes failing passes, ports new passes, reviews, and commits in a loop.
client-setup
Create a vanilla tRPC client with createTRPCClient (), configure link chain with httpBatchLink/httpLink, dynamic headers for auth, transformer on links (not client constructor). Infer types with inferRouterInputs and inferRouterOutputs. AbortController signal support. TRPCClientError typing.
adapter-express
Mount tRPC as Express middleware with createExpressMiddleware() from @trpc/server/adapters/express. Access Express req/res in createContext via CreateExpressContextOptions. Mount at a path prefix like app.use('/trpc', ...). Avoid global express.json() conflicting with tRPC body parsing for FormData.
trpc-router
Entry point for all tRPC skills. Decision tree routing by task: initTRPC.create(), t.router(), t.procedure, createTRPCClient, adapters, subscriptions, React Query, Next.js, links, middleware, validators, error handling, caching, FormData.
review-pending-pr-reviews
Review, grill, edit, and post pending PR review drafts saved by /review-pr (or its batch/cron runners). Lists drafts that have not yet been posted, walks the maintainer through each finding, then posts, closes, or discards. Use when the user says "post my pending reviews", "what reviews are waiting", "go through the…
compiler-port
Port a compiler pass from TypeScript to Rust. Gathers context, plans the port, implements in a subagent with test-fix loop, then reviews.