agentsclimarketplace
Sign in

Mega jobs

Skill ohtensorplay/mega-codex-plugin/plugins/mega/skills/mega-jobs

Plan, submit, inspect, monitor, cancel, and schedule MEGA container Jobs or operate MEGA Sandboxes and warm pools. Use for Job images, commands, arguments, timeouts, labels, logs, schedules, isolated execution, compute cost, output persistence, or deciding whether bounded MCP or the mega CLI is appropriate.From its SKILL.md

Install
npx -y skills add ohtensorplay/mega-codex-plugin --skill mega-jobs

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its file declares

Copied from the file, not written here

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

2.3 KB, 415 tokens by cl100k_base, as published. Nobody here has run it

MEGA Jobs

Use mega_jobs for bounded operations that contain no secrets. It supports list, inspect, logs, run, cancel, scheduled list/inspect/create/delete/suspend/resume/trigger.

Use mega_sandboxes to list providers, pools, and Sandboxes; create or delete warm pools; create or inspect an isolated Sandbox; collect metrics; run a bounded command; or stop a Sandbox. Repository mounts are read-only. Sandbox creation and execution can consume paid compute, and secret values are not an MCP input.

Plan before execution

Resolve:

  • Docker image and immutable version when possible.
  • Command, arguments, timeout, namespace, labels, and expected outputs.
  • Compute and billing implications.
  • Output persistence location and failure recovery.
  • Schedule, concurrency, and suspension behavior for recurring work.

Keep credentials and private keys out of MCP arguments. Route secret-bearing submissions to $mega-cli or the platform's protected secret workflow.

Execute safely

  1. Use list or inspect before canceling, deleting, suspending, resuming, or triggering a selected Job or schedule.
  2. Confirm the exact Job or schedule ID and explain the state change.
  3. Treat run and scheduled trigger as potentially paid operations.
  4. Bound log output with tail; use the CLI for streaming or long waits.
  5. Return the Job or schedule ID, current stage, request ID, and next monitoring action.

Do not report completion until the terminal stage and persisted outputs have been verified.

Sandbox safeguards

  1. Inspect providers or the selected Sandbox/pool before a lifecycle action.
  2. Confirm namespace, lifetime, egress choice, read-only mounts, command, working directory, timeout, and expected cost before creation or execution.
  3. Keep credentials, tokens, and private keys out of command arguments, stdin, and environment values. Use the protected platform workflow or CLI where appropriate.
  4. Use the CLI for interactive shells, streaming output, repeated waits, broad local mounts, and long-running automation.

What ships with it: 1 file

237 B alongside SKILL.md

agents/

Keep looking

Skills are one crate of 326,144. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.