Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add metaspartan/cybara --skill huggingface-trackiogit clone --depth 1 https://github.com/metaspartan/cybaraWrote 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/metaspartan/cybara/huggingface-trackio)<a href="https://agentmods.dev/skills/metaspartan/cybara/huggingface-trackio"><img src="https://agentmods.dev/badge/skills/metaspartan/cybara/huggingface-trackio/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/metaspartan/cybara/huggingface-trackio"><img src="https://agentmods.dev/badge/skills/metaspartan/cybara/huggingface-trackio.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00030 | $0.00322 |
| Opus 5 | $0.00015 | $0.00161 |
| Sonnet 5 | $0.00006 | $0.00064 |
| Haiku 4.5 | $0.00003 | $0.00032 |
Grade A, and why
huggingface-trackio 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 10d 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.
What it actually says
Hugging Face Trackio
Use Trackio to make training state observable and reproducible.
Instrumentation
import trackio
trackio.init(project="project-name", config={"learning_rate": 0.0001})
trackio.log({"loss": 0.1, "learning_rate": 0.0001, "step": 1})
trackio.finish()
Use report_to="trackio" when the selected trainer supports it. For remote training, configure a durable Space or other supported synchronization target so metrics survive the job.
Required signals
- train and evaluation loss
- task-specific evaluation metrics
- learning rate and step or epoch
- examples or tokens processed per second
- GPU memory or system utilization when available
- configuration, model revision, dataset revision, and seed
- alerts for NaN/Inf values, loss divergence, stalled progress, and failed persistence
Use structured CLI output when retrieving metrics for an agent. Compare runs only after verifying that their model, data, method, and evaluation settings are compatible.
Creating or changing a public tracking Space, webhook, or external alert destination transmits data. Confirm the destination and the metrics being sent before enabling it, and never log secrets or raw sensitive examples.
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.
- 10d ago First seen · 36 lines · 30 tokens per session scan A fb33f1ac4ce5
huggingface-trackio is a skill published in the GitHub repository metaspartan/cybara (28 stars, last pushed today), licensed MIT. It adds 30 tokens to every session and 322 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
kodama-verification
Define measurable success criteria and collect targeted test, build, lint, type-check, or smoke-test evidence before claiming work is complete.
kodama-behavior
Classify requests, assess unfamiliar codebases, delegate work safely, and recover from failed attempts. Use for multi-step, ambiguous, or parallelizable work.
kodama-constraints
Enforce non-negotiable safety, scope, security, and quality constraints for implementation and review work.
project-release-check
Validate this example project's release readiness. Use before publishing a release candidate.
anthropic-skill-development
Skill "anthropic-skill-development" from KornLabs/truss, covering skill development for claude code plugins, about skills, what skills provide, anatomy of a skill and progressive disclosure design principle.
marketing-ads
Skill "marketing-ads" from KornLabs/truss, covering paid ads, before starting, 1. campaign goals, 2. product & offer and 3. audience.