agentleFS
FracteraFractera/Agentic-Engineering-Infrastructure/.claude/skills/ultracode-session/SKILL.md

ultracode-session

Skill61 starsChanged 56 days agoMITView on GitHub

Build process. A session the person turned to ultracode in Telegram (Opus → Ultracode → the warning → «Продолжить»): the one place where the node allows dynamic workflows and subagents, and how it ends. Load it only when this session's system prompt says «Ultracode — this session only» or Claude Code's reminder says ultracode is on — never to switch it on yourself.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Only the person turns it on, in Telegram, while the agent sleeps: Opus → Ultracode → the warning about token use → «Продолжить».

    line 19

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

A confirmed ultracode session

Process: build — an ultracode session is a build session: it ends with a change in git, like any other.

A hint, not a law: models of late 2026 work worse under rigid step-by-step instructions. The laws below are not hints.

What it is (Claude Code 2.1.319, read in its own code — node step 666)

  • Ultracode is standing dynamic-workflow orchestration at any effort level: you write a workflow script, Claude Code's runtime runs it, and its agents run on this computer with this session's tools and permissions — their file edits are auto-approved.
  • The node turns it on for one session: --model opus --effort max --settings {"enableWorkflows":true,"ultracode":true}, and a folder with its own tool list (.claude/launch-tools.txt) gets Agent,Workflow. Every other session of the node starts with workflows off, ultracode off and an explicit effort — medium unless the person chose another (bridge.cjs, 666).
  • Only the person turns it on, in Telegram, while the agent sleeps: Opus → Ultracode → the warning about token use → «Продолжить». Never you: no /effort, no /config, no word «ultracode» in a prompt or in a task for another agent.

What changes here

  • The ban on subagents (CLAUDE.md §6) is lifted in this session and only in it. Every other law stands: the element boundary, git, the person's «yes» before publishing, two proofs before «done».
  • A workflow pays where every unit is judged by a machine without the person — every dictionary, every content page, one sweep across all routes; read .claude/skills/use-dynamic-workflows/SKILL.md by path for the rule and the failure mode. Architecture, one stubborn build, a design decision — one mind holding the whole picture is better; tell the person so instead of fanning out.
  • Name the guard before the wave: the command that judges each unit (check:i18n, check:content, the build).
  • Name the price before a large wave and run one slice first. The size guideline is small (<5 agents) by default on Pro, medium (<10) elsewhere, large <50; how many run at once is Claude Code's limit (CLAUDE_CODE_MAX_CONCURRENT_SUBAGENTS).
  • Telegram shows only «запускаю рабочий процесс агентов» / «передаю часть субагенту» in the thinking lines — never the agents' texts.

The end

  • «выключить» in any words → npm run done -- --close --note "…", as always. The node stops the session, writes the reset (data/services/build/session-mode.json: medium, workflows off) and sends its own closing words: «Агент завершил работу, контекст сброшен, усилия восстановлены в положении Medium, рабочие процессы отключены». Do not repeat them and do not promise ultracode next time — the person chooses it again.
  • A model switch in the middle (npm run session -- model …) continues the conversation at medium without workflows: ultracode does not survive it.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.