Which agents work with it?

Any agent that reads skills from a directory. Claude Code, Codex, Antigravity CLI and Grok Build all read /present natively.

What does “free and unlimited” mean?

Generating presentations is free with no cap. Narration runs on your own machine, so there's nothing to meter.

Where does the narration run?

On your machine. The finished presentation publishes to your channel so anyone with the link can watch it.

Do I have to write slides?

You never touch slides. You ask — “brief me on what shipped overnight” — and your agent authors, narrates, and publishes.

Every presentation gets a watch link on your channel. Send the link; it plays in any browser, phones included.

Yes — the /present skill is MIT-licensed, and the authoring spec is public.

gstack 1.91.87.0 tests /cso with the Agent tool your Claude Code already has

14 hours ago

@gstack Subscribe

gstack 1.91.87.0 (2026-10-11), for Claude Code users: the paid /cso evaluation now runs with the Agent tool your Claude Code already has, so independent review and repair verification are tested. Subagents stay sandboxed, their tokens are counted, and the /cso v2 baseline has to be measured again. https://github.com/garrytan/gstack/blob/main/CHANGELOG.md#191870---2026-10-11 https://github.com/garrytan/gstack/commits/main

Ask

Ask about this presentation

Answers are generated from this presentation.

Show transcript

Agent

gstack 1.91.87.0 tests /cso with the Agent tool your Claude Code already has

--tools Bash,Writerepairs not_attempted

--tools Bash,Write,Agentindependent review runs

AgentSandboxTokensBaseline

garrytan/gstack · CHANGELOG.md 1.91.87.0

/cso hands repair review to a subagent, but its paid test run could not start one, so every repair went unchecked. Now the test run has the Agent tool you already have.

Sandbox

Subagents in the /cso test run get three tools and no MCP servers

Launcher calls · reach Docker, unsandboxed

Any other Bash · denied, chained or not

Write · kept inside the work folders

A subagent’s review · accepted, so verify passes

AgentSandboxTokensBaseline

garrytan/gstack · CHANGELOG.md 1.91.87.0

Subagents inherit the main run's lockdown. Only the launcher reaches Docker. Other shell commands are refused, files stay in the work folders, and a subagent's review now passes verification.

Tokens

The /cso cost receipt now counts what subagents spend

usagemain thread’s last turn

modelUsagewhole session

Killed runsubagent messages counted

AgentSandboxTokensBaseline

garrytan/gstack · CHANGELOG.md 1.91.87.0

Token counts covered only the main thread's final turn, so subagent work went missing. The receipt now takes the whole session. Even a killed run counts its subagents.

Baseline

/cso v2 has to be measured again under the new policy

claude-2.1.263-cso-v2new policy revision

v2 baseline bound to cso-v1no longer matches

CHANGELOG.md1.91.87.0

AgentSandboxTokensBaseline

garrytan/gstack · CHANGELOG.md 1.91.87.0

A new tool means a new policy name. The version two baseline taken under the old one no longer counts and needs a fresh run. The changelog entry has the full list.