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 chipi/agentic-ai-homelab --skill fleet-statsgit clone --depth 1 https://github.com/chipi/agentic-ai-homelabWrote 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/chipi/agentic-ai-homelab/fleet-stats)<a href="https://agentmods.dev/skills/chipi/agentic-ai-homelab/fleet-stats"><img src="https://agentmods.dev/badge/skills/chipi/agentic-ai-homelab/fleet-stats/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/chipi/agentic-ai-homelab/fleet-stats"><img src="https://agentmods.dev/badge/skills/chipi/agentic-ai-homelab/fleet-stats.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00078 | $0.00753 |
| Opus 5 | $0.00039 | $0.00377 |
| Sonnet 5 | $0.00016 | $0.00151 |
| Haiku 4.5 | $0.00008 | $0.00075 |
Grade A, and why
fleet-stats 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 9d 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 — 61 lines — stays where its author put it; the contents beside it link to each section on GitHub.
fleet-stats
Read the current session's transcript and report how the subagent fleet behaved — who ran, how often, tokens per agent and per model tier, and how often executors escalated to the advisor. Answers "does the cost-spread work" from real data. Local + read-only.
Run
python3 ~/.claude/skills/fleet-stats/scripts/fleet_stats.py # latest transcript for this cwd
# or a specific transcript:
python3 ~/.claude/skills/fleet-stats/scripts/fleet_stats.py <path-to-session>.jsonl
It parses each toolUseResult with an agentType (a top-level subagent
invocation) for totalTokens + resolvedModel, then walks
<session>/subagents/ for nested escalations (see "How nesting is counted").
Read the report
- invocations per agent — who ran, how many times, on which model, tokens.
- token + cost spread by model tier — the money shot:
haiku/sonnetshould carry the volume;opusshould be a small slice (advisor escalations only). Ifopusdominates, the tiering isn't paying off. - escalation — nested executor→advisor calls.
- delegation diagram — mermaid; who called whom.
Cost rates in the script are approximate — update to current pricing (see the
claude-api skill); tokens-per-model is the real signal.
How nesting is counted (v2)
Only top-level subagent calls (those the orchestrator spawned) appear in the
main transcript. A nested call — an executor consulting the advisor — runs
as a separate sub-agent, recorded in <session>/subagents/agent-<id>.jsonl, and
never lands in the main transcript. v1 read only the main transcript, so nested
escalations were invisible: escalation showed 0 nested and the advisor's
(opus) tokens went uncounted entirely (they do not roll into the parent — a
parent's totalTokens is its own final-context, disjoint from the child's).
v2 walks the subagents/ dir too:
- Top-level rows use the authoritative
totalTokensfrom the main transcript — identical numbers to v1. - Nested rows read the sub-agent's own final-context (input + cache + output
of its last turn — exactly what
totalTokensmeasures) from its transcript, and link back to the parent that spawned it viameta.toolUseId. Some escalations (e.g. a reviewer's advisor consult) are never persisted as a result in the parent, so the sub-transcript is the only source.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 9d ago First seen · 61 lines · 78 tokens per session scan A 3f5aca071c86
fleet-stats is a skill published in the GitHub repository chipi/agentic-ai-homelab (2 stars, last pushed 2d ago), licensed MIT. It adds 78 tokens to every session and 753 once invoked, about $0.0004 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 2026-08-31.
Other skills, from other repositories
mikromcp
Use when managing, configuring, or troubleshooting MikroTik RouterOS devices through the MikroMCP server — picking the right tool, running safe (dry-run, confirmed, rollback-able) changes, fleet operations, and diagnosing network issues.
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.
jetson-diagnostic
Read-only Jetson health snapshot for identity, memory, GPU, thermal, power, storage, services, and top processes.
nemo-automodel-launcher-config
Configure NeMo AutoModel job launches for interactive runs, Slurm clusters, and SkyPilot cloud execution.
dogfood
Systematically explore and test a mobile app on iOS/Android with agent-device to find bugs, UX issues, and other problems. Use when asked to dogfood, QA, exploratory test, find issues, bug hunt, or test this app on mobile.
mochi-remind
Handle due reminders — notify the user with natural language and mark them done.