New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Rl Agentic Launch Jupiter

by open-thoughts in open-thoughts/OpenThoughts-Agent

Launch / relaunch agentic RL (SkyRL terminalbench + Harbor + Daytona) on JSC Jupiter (GH200).

Apache-2.0Auto-check passed

Install Rl Agentic Launch Jupiter

skills CLI
$ npx skills add open-thoughts/OpenThoughts-Agent --skill rl-agentic-launch-jupiter -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install open-thoughts/OpenThoughts-Agent rl-agentic-launch-jupiter --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/open-thoughts/OpenThoughts-Agent.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.agents/skills/rl-agentic-launch-jupiter .claude/skills/rl-agentic-launch-jupiter && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
rl-agentic-launch-jupiter
GitHub stars
301
Token cost
~2.7k tokens
SKILL.md length
1,077 words
Files
1
Skills in repo
44
Repo updated
First seen
Licence
Apache-2.0

At a glance

Launch / relaunch agentic RL (SkyRL terminalbench + Harbor + Daytona) on JSC Jupiter (GH200).

  • Works in 7 steps: The canonical launch → Config map + node count (num_nodes MUST… → Runtime / SIF selection → …
  • Asked to launch / relaunch / refill an agentic SkyRL RL run on Jupiter
  • SKILL.md covers 1. The canonical launch, 2. Config map + node count…, 3. Runtime / SIF selection and 4. Agentic infra conventions, plus 3 more sections
  • Calls python and git; needs DAYTONA_RL_API_KEY

What it does

Rl Agentic Launch Jupiter is an agent skill from open-thoughts/OpenThoughts-Agent. Launch / relaunch agentic RL (SkyRL terminalbench + Harbor + Daytona) on JSC Jupiter (GH200). Covers the dense 8B/32B FSDP2 arms (seqnorm, TIS, shaped, symclip, lrboost, loopshape) and the MoE/80B Megatron arms (Qwen3-Coder-30B-A3B, Qwen3-Next-80B-A3B) — the exact python -m hpc.launch --jobtype rl flag set, which flags vary per arm (config / modelpath / traindata / numnodes), runtime+SIF selection, the Daytona RL-org + chain-restart conventions, and the standing constraints (≤6 RL/cluster, a3 CONCLUDED, TIMEOUT…

Its SKILL.md is about 2.7k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It works with Qwen, NVIDIA AI Platform and Python. The repository describes itself as: Data recipes and robust infrastructure for training AI agents. The licence is Apache-2.0.

When your agent uses it

  • Asked to launch / relaunch / refill an agentic SkyRL RL run on Jupiter

Example prompts

  • “/rl-agentic-launch-jupiter”

Requirements

  • Python 3
  • A credential in DAYTONA_RL_API_KEY

Workflow steps

7 steps, taken from the step headings in SKILL.md.

  1. The canonical launch
  2. Config map + node count (num_nodes MUST match the config)
  3. Runtime / SIF selection
  4. Agentic infra conventions
  5. Chain-restart (--max_restarts K)
  6. Standing constraints (do NOT violate)
  7. After launch

What it can do on your machine

Read from SKILL.md and the folder at commit 3bd1917. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • python
    • git

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use git, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names these keys or tokens, usually read from environment variables:

    • DAYTONA_RL_API_KEY

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Rl Agentic Launch Jupiter loads about 2.7k tokens when it runs. Until then it costs about 187 tokens; SKILL.md has 1,077 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~187
When it runs · the whole SKILL.md, loaded when a task matches
~2.7k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from open-thoughts/OpenThoughts-Agent at commit 3bd1917, republished under its Apache-2.0 licence (© open-thoughts). 1,077 words, ~2,715 tokens.

Download SKILL.mdSave it as .claude/skills/rl-agentic-launch-jupiter/SKILL.md (or your agent's skills folder).
name
rl-agentic-launch-jupiter
description
Launch / relaunch agentic RL (SkyRL terminal_bench + Harbor + Daytona) on JSC Jupiter (GH200). Covers the dense 8B/32B FSDP2 arms (seqnorm, TIS, shaped, symclip, lrboost, loopshape) and the MoE/80B Megatron arms (Qwen3-Coder-30B-A3B, Qwen3-Next-80B-A3B) — the exact `python -m hpc.launch --job_type rl` flag set, which flags vary per arm (config / model_path / train_data / num_nodes), runtime+SIF selection, the Daytona RL-org + chain-restart conventions, and the standing constraints (≤6 RL/cluster, a3 CONCLUDED, TIMEOUT restarts are normal). Use when asked to launch / relaunch / refill an agentic SkyRL RL run on Jupiter. Reference: notes/ot-agent/rl_experiments.md, .agents/ops/jupiter/{ops.md,ENVIRONMENT_MAP.md}.

⚠ Do not add comments to YAMLs. Report your recommendations directly to the supervisor.

rl-agentic-launch-jupiter

⚠ Local clone = ground truth (CLAUDE.md §Always). ALL code/config/sbatch edits (OpenThoughts-Agent + MarinSkyRL) go in the local Mac checkouts → commit → push → git pull on the cluster. NEVER hand-edit, git commit, or leave divergent/ untracked changes on a cluster; no patch-by-rsync (vLLM is the only exception — built from source per-cluster). Bake this into every subagent you dispatch.

Agentic SkyRL/GRPO RL runs through python -m hpc.launch --job_type rl with FSDP2 or Megatron. Each rollout is a Harbor agent episode in a Daytona terminal_bench sandbox with a colocated vLLM engine. Jupiter nodes have four 96GB GH200 GPUs. Read .agents/ops/jupiter/ops.md first; runtime/SIF details are in .agents/ops/jupiter/ENVIRONMENT_MAP.md.

1. The canonical launch

🚧 SUBMIT FROM THE REPO DIR WITH DCFT SET. Before launching/resuming: cd /e/scratch/jureap59/feuer1/OpenThoughts-Agent && export DCFT=$PWD (the ops.md preamble does this). The generated universal_rl.sbatch resolves WORKDIR from DCFT_PRIVATE → DCFT → $PWD; submitted from $HOME/a scratch subdir with DCFT unset, the guard detects the wrong dir (missing hpc/shell_utils/triton_cache.sh marker) and exit 1s with FATAL: WORKDIR=... is not the OpenThoughts-Agent repo root. Fix: cd to the repo, export DCFT=$PWD, resubmit.

bash
python -m hpc.launch --job_type rl \
  --rl_config ./hpc/skyrl_yaml/jupiter/<cfg>.yaml \
  --model_path <hf-or-local-model> \
  --train_data '["<HF-repo-or-/abs/task/dir>"]' \
  --num_nodes N \
  --time_limit 11:59:00 \
  --max_restarts K \
  --reservation reformo \
  --experiments_dir /e/data1/datasets/playground/ot-baf \
  --job_name <name>

Varies per arm: --rl_config, --model_path, --train_data, and --num_nodes (§2). Fixed on Jupiter:

  • --time_limit 11:59:00 — booster QOS caps walltime at 12h; chain with --max_restarts (§5).
  • --reservation reformo — jureap59 booster QOS is suspended (InvalidQOS); reformo is the runnable account/reservation.
  • --experiments_dir /e/data1/datasets/playground/ot-baf — the ot-baf personal data root (/ot is read-only-for-you).
  • --train_data is a JSON-list string '["..."]' — an HF repo (DCAgent/…, laion/…, SankalpKJ/…) or a pre-extracted local task dir (/e/scratch/jureap59/feuer1/tasks/<name>).
  • --job_name <name> — set explicitly for predictable chain-restart and cleanup paths.
  • --skyrl_override '++a.b.c=val' — appends a Hydra override (last-wins over the base yaml). For per-arm tweaks without forking a config: sampling (generator.sampling_params.temperature=1.0, …top_p, …top_k, …min_p), Harbor sandbox sizing (++terminal_bench_config.harbor.override_{cpus,memory_mb,storage_mb}), context bumps (++generator.engine_init_kwargs.max_model_len=…). Pass ++-prefixed, struct-safe keys — a bare top-level key risks a Hydra ConfigKeyError.
  • Launch from the otagent conda env (/e/scratch/jureap59/feuer1/miniforge3/envs/otagent/bin/python), NOT the RL venv — task extraction imports google.cloud.storage, which the RL venv lacks. (The launcher then selects the RL venv/SIF for the training — §3.)

2. Config map + node count (num_nodes MUST match the config)

num_nodes = GPUs / 4. Pick the config, then set --num_nodes to its budget:

Config (hpc/skyrl_yaml/jupiter/…)ModelGPUs → --num_nodes
56GPU_seqnorm_tis.yaml (+ extra/56GPU_seqnorm.yaml, extra/56GPU_seqnorm_tis_shaped.yaml)dense 8B56 → 14
extra/56GPU_seqnorm_tis_untrunc_symclip.yamldense 8B (symclip)56 → 14
extra/56GPU_seqnorm_tis_untrunc_symclip_loopshape.yamldense 8B (symclip+loopshape)56 → 14
extra/56GPU_seqnorm_tis_untrunc_lrboost.yamldense 8B (lr-boost)56 → 14
56GPU_shaped.yaml (extra/24GPU_shaped.yaml)dense 8B (shaped reward)56→14 / 24→6
24GPU_base_131k.yaml / extra/24GPU_base_old.yamldense 8B24 → 6
64GPU_base_32b.yaml, extra/64GPU_base_32b_fp8.yaml, extra/48GPU_*_32b.yaml, extra/128GPU_base_32b.yamldense 32B64→16 / 48→12 / 128→32
24GPU_qwen3_coder_30b_a3b.yamlQwen3-Coder-30B-A3B (MoE)24 → 6
extra/128GPU_qwen3_next_80b_a3b.yamlQwen3-Next-80B-A3B (MoE, prod)64 → 16 (name is historical; header = 64 GPU/16 node)
extra/16GPU_mixtral_8x7b.yamlMixtral-8x7B (MoE bring-up)16 → 4

General rule: 24GPU→6, 48GPU→12, 56GPU→14, 64GPU→16, 96GPU→24, 128GPU→32. The CLI controls -N despite generated #SBATCH --nodes=1. For an unexplained <15-minute failure, check node count first.

3. Runtime / SIF selection

The launcher selects the training runtime (hpc/sbatch_rl/universal_rl.sbatch). Confirm it from the rendered sbatch, rather than assuming.

  • Dense 8B/32B FSDP2 (seqnorm / TIS / shaped / symclip / lrboost / loopshape) → RL venv $WORKDIR/envs/rl (torch 2.9). Default RL runtime.
  • MoE — Qwen3-Coder-30B-A3B and prod 80B Qwen3-Next-80B-A3B (R3+TIS) → SIF skyrl_megatron_vllm_r3baked.sif (torch 2.9, overlays baked in).
  • torch≥2.10 / DCP / torch-native CP / Mixtral-multinode → SIF skyrl_megatron_vllm0202rc0_r3.sif (torch 2.11); stack the skyrl_titan_overlay.img when torchtitan-0.2.2 / _StridedShard (CP+EP) is needed.

Use torch, not vllm.__version__, to identify the runtime. See ENVIRONMENT_MAP §4 for probes and SIF gotchas.

Show full SKILL.md (537 more words)Show less

4. Agentic infra conventions

  • Daytona uses the RL-org key for RL rollouts (distinct from the eval-org key); set by the launch preamble / hpc/dotenv/jupiter.env, not the CLI.
  • Pinggy is EVAL-only, not RL — --pinggy_persistent_url / --pinggy_token are eval-path flags.
  • enable_db_registration: false — the launcher auto-injects ++trainer.enable_db_registration=false for RL. Do NOT also pass a bare --skyrl_override enable_db_registration=false (Hydra struct ConfigKeyError risk, redundant). DB registration is a manual cleanup step, not a launch flag.
  • Daytona snapshots: a new task set builds snapshots on first launch; caps are HARD (10 new / 60 org). At the org cap, clean stale snapshots first; do not raise the cap: python scripts/daytona/daytona_snapshot_manager.py --api-key-env DAYTONA_RL_API_KEY --delete-stale --yes (deletes only idle/unprotected harbor__* envs — safe; threshold in .agents/projects/daytona/daytona.md). Only a single dataset legitimately needing

    max_new_snapshots unique envs escalates → ask.

  • vLLM DP>1 (ray backend): never hardcode --data-parallel-address 127.0.0.1 — Ray registers the head only under its real IPv4 → 127.0.0.1 gives AssertionError: DP master node missing or dead. hpc/vllm_utils.py VLLMServer.start() auto-injects the head IP for DP>1; don't add the flag to new yamls. If overriding, use the real Ray head IPv4.
  • MoE / 80B placement: the MoE configs carry their own FSDP/EP sizing in-yaml (Coder-30B: EP=4×FSDP=4=16 policy GPUs + 4 TP=2 vLLM engines = 24 GPU/6 nodes; 80B: 8 TP=4 engines + 8-node FSDP shard = 64 GPU/16 nodes). The 80B yaml sets policy_strict_spread_pg: true (opt-in anti-affinity reserving the policy PG up front to dodge the two-PACK-PG init-OOM race); leave as-configured. Honor the MoE FSDP/EP divisibility constraint (fsdp_size must divide num_experts // ep_size) — don't hand-edit node/EP counts. Details → .agents/projects/marinskyrl/marinskyrl.md.

5. Chain-restart (--max_restarts K)

--max_restarts K submits a head job + K afterany-dependent restart links. A link that hits the 12h wall TIMEOUT auto-resumes from the latest checkpoint in the next link — TIMEOUT is the NORMAL terminal state of a healthy chain, not a failure. Typical K = 5–6.

  • A fresh python -m hpc.launch with the SAME --job_name forks to <dir>_2 at step 0 if the original exp dir's configs/*.json exists (the dedup resume-manager engages only for datagen/eval, not RL). To resume instead of forking: either resubmit the existing generated sbatch (experiments/<dir>/sbatch/*_rl.sbatch) via --dependency=afterany, or move the original configs/*.json aside so dedup lands on the un-suffixed dir. (--dry_run regenerates that config → re-move after a dry-run, or skip it.)
  • Relaunching auto-resumes from checkpoints/global_step_N/. For a clean ablation, remove <exp>/<job>/<job>/checkpoints/ before relaunching; retain it for a chain extension.
  • Always scancel the previous failed/superseded chain before resubmitting.

6. Standing constraints (do NOT violate)

  • Daytona RL concurrency ≤ 6 RUNNING per cluster (PENDING restart links don't count). Don't launch a 7th concurrent RL job on Jupiter.
  • The a3 series is CONCLUDED — do NOT launch, refill, or auto-advance a3 rows (binary reward + RLOO-n + token_mean; uninformative). Successor arms = the seqnorm / TIS / shaped / symclip / loopshape ablations above. (Exception: DCAgent/r2egym-patched-full-oracle is a separate snapshot-optimized variant — not the a3 row — and launches fine.)
  • Never alter config/hparams mid-series. A controlled ablation needing a change → propose a separate experiment; don't mutate the in-flight arm.
  • TIMEOUT restarts are expected (§5) — don't treat a chain's TIMEOUT links as failures or salvage them.

7. After launch

  • Monitor: monitor-cron-sweep (entropy / log_ratio / grad_norm are mandatory progress columns).
  • On completion → rl-agentic-job-cleanup (best-ckpt selection, HF upload from the login node, the manual Supabase DB registration, trace export + parse_skyrl_metrics). enable_db_registration stays false at launch (§4).
  • Behavior analysis: analyze-rl-behavior for a post-hoc arm comparison.

© open-thoughts, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .agents/skills/rl-agentic-launch-jupiter of open-thoughts/OpenThoughts-Agent.

Open the folder on GitHubat commit 3bd1917

Compare with similar skills

Rl Agentic Launch Jupiter next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Rl Agentic Launch Jupiter compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Rl Agentic Launch Jupiter this skillopen-thoughts/OpenThoughts-Agent301—~2.7kAutomated safety check: PassApache-2.0
AI Search Hubminsight-ai-info/AI-Search-Hub1.3k—~1.3kAutomated safety check: PassNone
Qwen Mtp GgufR6410418/Jackrong-llm-finetuning-guide1.7k—~1.7kAutomated safety check: PassMIT
Refactor OpCVCUDA/CV-CUDA2.7k—~1.5kAutomated safety check: PassCustom licence
Dstack Prototypingdstackai/dstack2.3k—~1.6kAutomated safety check: PassMPL-2.0
Pocketmen With Yousix-nut/PocketMen-with-you310—~2.6kAutomated safety check: PassMIT

Similar skills

  • AI Search Hub

    minsight-ai-info/AI-Search-Hub

    Run the AI Search Hub browser automation scripts for Yuanbao, LongCat, Doubao, Qwen, Gemini, Grok, and MiniMax.

    1.3k GitHub stars~1.3k tokensUpdated 5 mo ago
    Productivity & AutomationAuto-check passed

  • Qwen Mtp Gguf

    R6410418/Jackrong-llm-finetuning-guide

    Complete agent-ready workflow for Qwen-family MTP or nextn GGUF conversion and release.

    1.7k GitHub stars~1.7k tokensUpdated 2 mo ago
    AI & LLM EngineeringAuto-check passed

  • Refactor Op

    CVCUDA/CV-CUDA

    Find and safely apply per-operator refactoring / redundancy-reduction opportunities in a CV-CUDA operator (near-duplicate Tensor/VarShape kernels, reinvented shared utilities, dead code).

    2.7k GitHub stars~1.5k tokensUpdated 22 days ago
    DevelopmentAuto-check passed

  • Dstack Prototyping

    dstackai/dstack

    Use with the dstack skill for model-serving work when the image, serving command, resources, backend/fleet choice, or service behavior is not proven.

    2.3k GitHub stars~1.6k tokensUpdated yesterday
    AI & LLM EngineeringAuto-check passed

  • Pocketmen With You

    six-nut/PocketMen-with-you

    Turn 2+ user reference images into a high-fidelity animated Codex companion using PocketMen's own local stack.

    310 GitHub stars~2.6k tokensUpdated 7 days ago
    AI & LLM EngineeringAuto-check passed

  • Gds Diag

    NVIDIA/MagnumIO

    Official

    A skill your agent uses when diagnosing NVIDIA GPUDirect Storage with this repository: choose and run the right gds-diag.py subcommand, interpret its output, and explain operator next steps without…

    125 GitHub stars~1.6k tokensUpdated yesterday
    Auto-check passed

More from open-thoughts/OpenThoughts-Agent

All 44 skills in this repo

  • Analyze Dataset Token Length

    open-thoughts/OpenThoughts-Agent

    Analyze the token length of an OT-Agent conversation-format (ShareGPT-style) dataset — the per-trace distribution (median/p90/max) and/or counts under a token threshold + a metadata predicate (e.g.

    301 GitHub stars~1.5k tokensUpdated 10 days ago
    Auto-check passed

  • Analyze Id Eval Ranking

    open-thoughts/OpenThoughts-Agent

    Given a list of models (HF name stubs) that have valid agentic ID eval scores in Supabase, build a ranking table: raw per-benchmark accuracy on the 3 ID benchmarks (SWE-Bench-100…

    301 GitHub stars~3.1k tokensUpdated 10 days ago
    Auto-check passed

  • Analyze Job History Iris

    open-thoughts/OpenThoughts-Agent

    Run the Iris harbor job-history analyzer (scripts/iris/analyzeirisharborjob.py) on a datagen/eval job and read its JSON sidecar for trustworthy throughput / preemption / productive-trial stats.

    301 GitHub stars~2.9k tokensUpdated 10 days ago
    Auto-check passed

  • Analyze Rl Behavior

    open-thoughts/OpenThoughts-Agent

    Run the full RL behavioral-analysis pipeline (scripts/analysis/analyzerlbehavior.py) on a trained RL model to understand WHAT changed vs its pre-RL baseline, WHY, whether it PERSISTS, and its EVAL…

    301 GitHub stars~4.2k tokensUpdated 10 days ago
    Auto-check passed

  • Analyze Training Run Iris

    open-thoughts/OpenThoughts-Agent

    Detailed health check for a Levanter/executor TRAINING run on the marin Iris cluster (e.g.

    301 GitHub stars~2k tokensUpdated 10 days ago
    Auto-check passed

  • Code Create Staged Plan

    open-thoughts/OpenThoughts-Agent

    DESIGN a non-trivial codebase change (Harbor / MarinSkyRL / vLLM / OT-Agent / LLaMA-Factory) as a dependency-ordered STAGED PLAN before writing code — a feature port, a multi-step fix with parity…

    301 GitHub stars~1.5k tokensUpdated 10 days ago
    Auto-check passed

Works with

Questions about Rl Agentic Launch Jupiter

What does Rl Agentic Launch Jupiter do?

Launch / relaunch agentic RL (SkyRL terminalbench + Harbor + Daytona) on JSC Jupiter (GH200). Rl Agentic Launch Jupiter is an agent skill from open-thoughts/OpenThoughts-Agent. Launch / relaunch agentic RL (SkyRL terminalbench + Harbor + Daytona) on JSC Jupiter (GH200).

When should I use Rl Agentic Launch Jupiter?

Rl Agentic Launch Jupiter fits situations like: asked to launch / relaunch / refill an agentic SkyRL RL run on Jupiter.

How do I install Rl Agentic Launch Jupiter in Claude Code?

Run `npx skills add open-thoughts/OpenThoughts-Agent --skill rl-agentic-launch-jupiter -a claude-code`. Or copy the skill folder (.agents/skills/rl-agentic-launch-jupiter in open-thoughts/OpenThoughts-Agent) into .claude/skills/rl-agentic-launch-jupiter in your project. Claude Code loads it when a task matches its description.

How do I install Rl Agentic Launch Jupiter in Codex?

Run `npx skills add open-thoughts/OpenThoughts-Agent --skill rl-agentic-launch-jupiter -a codex`. Or copy the skill folder (.agents/skills/rl-agentic-launch-jupiter in open-thoughts/OpenThoughts-Agent) into .agents/skills/rl-agentic-launch-jupiter in your project. Codex loads it when a task matches its description.

Can I use Rl Agentic Launch Jupiter in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add open-thoughts/OpenThoughts-Agent --skill rl-agentic-launch-jupiter -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/rl-agentic-launch-jupiter, .gemini/skills/rl-agentic-launch-jupiter, .github/skills/rl-agentic-launch-jupiter and .opencode/skills/rl-agentic-launch-jupiter in your project.

What does Rl Agentic Launch Jupiter need to run?

Going by SKILL.md and its folder, Rl Agentic Launch Jupiter needs the command-line tools its instructions call (python and git) and credentials named DAYTONA_RL_API_KEY. Our summary lists: Python 3; A credential in DAYTONA_RL_API_KEY.

Does Rl Agentic Launch Jupiter access the network?

SKILL.md contains no URLs. Its commands use git, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Rl Agentic Launch Jupiter safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Rl Agentic Launch Jupiter use?

Rl Agentic Launch Jupiter is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Rl Agentic Launch Jupiter use?

About 2.7k tokens (SKILL.md is roughly 11k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Rl Agentic Launch Jupiter?

Skills that share tags, products or a category with Rl Agentic Launch Jupiter: AI Search Hub (minsight-ai-info/AI-Search-Hub, 1.3k stars), Qwen Mtp Gguf (R6410418/Jackrong-llm-finetuning-guide, 1.7k stars), Refactor Op (CVCUDA/CV-CUDA, 2.7k stars) and Dstack Prototyping (dstackai/dstack, 2.3k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Rl Agentic Launch Jupiter?

open-thoughts (a GitHub organization) maintains it in open-thoughts/OpenThoughts-Agent, which has 301 GitHub stars. The repository holds 44 skills in this directory. The repository was last updated on September 28, 2026.

Source: open-thoughts/OpenThoughts-Agent on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.