Marin is an open-source research program, software platform, and community for developing foundation models such as large language models. Researchers use it for data preparation, tokenization, pretraining, posttraining, evaluation, and related experiments, including work on audio-text, DNA, and protein models. The catalogue entries are add-ons that support workflows around Marin.
Borrowing it
Nothing to install: this file belongs to marin-community/marin. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/marin-community/marin/main/.agents/skills/manage-hero-run/SKILL.mdgit clone --depth 1 https://github.com/marin-community/marinWrote 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/marin-community/marin/manage-hero-run)<a href="https://agentmods.dev/skills/marin-community/marin/manage-hero-run"><img src="https://agentmods.dev/badge/skills/marin-community/marin/manage-hero-run/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/marin-community/marin/manage-hero-run"><img src="https://agentmods.dev/badge/skills/marin-community/marin/manage-hero-run.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Excessive Agency · line 115 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
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 | $0.00034 | $0.03164 |
| Opus 5 | $0.00017 | $0.01582 |
| Sonnet 5 | $0.00007 | $0.00633 |
| Haiku 4.5 | $0.00003 | $0.00316 |
Grade A, and why
manage-hero-run scanned grade A with 0 findings 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 12d 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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 174 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Manage Hero Run
Use this skill only when the request names a hero or production-critical run, or another selected workflow explicitly delegates its management here. Within this workflow, about 1e22 model FLOPs (using a 6ND estimate) is the normal threshold for classifying a run as production rather than bounded diagnostic work; it is not an activation trigger by itself.
Operating Model
- If the user asks to launch a run, arrange babysitting unless they explicitly decline.
- Use a 15 minute default check cadence for babysitting unless the run's failure mode requires tighter monitoring.
- Never stop, restart, or bounce an Iris cluster without explicit user permission.
- If something needs human judgment or authorization, attempt to contact the DRI, usually the user in the chat, through available channels such as GitHub issue comments, Discord, email, or Slack.
Classify the run before loading supporting workflows:
- A production run is long-lived, has a durable output contract, or is expected to exceed the 1e22 FLOP guideline. It requires the full run record below.
- A bounded diagnostic has a small fixed step or time limit, lifecycle-managed output, no canonical export, and one monitoring owner. Record its launch contract in the originating conversation or durable session channel. Do not create an issue or logbook solely to submit it.
Do not load run-research, task-logbook, task-snapshot, file-issue, or their writing guides
for a bounded diagnostic. Load them only when the run requires the artifact they govern.
Launch
Start from the user's reference command or the nearest existing launcher. Read that launcher's
README and --help, then apply only the requested differences. Do not reconstruct the experiment
from implementation internals or add a launcher when existing flags express the run.
Resolve the launch blockers first:
- Confirm the requested source commit is present with a direct ancestry check.
- Inspect the reference job read-only for the fields that the new run must preserve.
- Choose unique job, run, output, checkpoint, and rendezvous identities.
- Print the dry plan and exact submit command.
- Submit once the checks below are resolved. Issue setup, prose polishing, broad repository tests, and prior-work searches do not block a bounded diagnostic.
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.
- 12d ago First seen · 174 lines · 34 tokens per session scan A c6f4081d65da
manage-hero-run is a skill published in the GitHub repository marin-community/marin (3,593 stars, last pushed yesterday), licensed Apache-2.0. It adds 34 tokens to every session and 3,164 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
agent-platform-rag-engine-management
Manage and query Agent Platform RAG Engine Corpora and retrieve grounded contexts using the Google GenAI SDK. Use when listing RAG corpora or files, inspecting a corpus, retrieving contexts, or generating content grounded in a RAG corpus. Do not use for standard database queries (use SQL/Spanner skills), Google…
agent-platform-model-registry
Agent Platform Model Registry Management. Use when you need to upload, list, describe, update, or delete machine learning models (and their versions) in the Agent Platform Model Registry. Don't use for model training, model deployment to endpoints, or managing non-Agent Platform models.
foundry-config-setup
Resolve missing setup caused by a hardcoded Foundry project endpoint or model in a sample. Use when a sample fails because it uses a placeholder/hardcoded projectendpoint (for example "https://your-project.services.ai.azure.com") or a hardcoded model instead of reading them from the environment.
google-cloud-solution-agentic-analytics-spark-knowledge-catalog
Discovers requirements and generates guidance to design and deploy a governed, secure agentic-analytics solution for data that's distributed across Google Cloud, other cloud providers, or on-premises. Data that's outside Google Cloud (such as data from Databricks, Snowflake, Salesforce, SAP, or Oracle systems) is…
training-check
Interactively monitor training metrics from the current Codex session, periodically checking WandB or fallback logs for NaN, divergence, plateaus, and broken runs.
nemo-automodel-launcher-config
Configure NeMo AutoModel job launches for interactive runs, Slurm clusters, and SkyPilot cloud execution.