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 santoshkanthety/powerbi-agent --skill powerbi-fabric-capacitygit clone --depth 1 https://github.com/santoshkanthety/powerbi-agentWrote 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/santoshkanthety/powerbi-agent/powerbi-fabric-capacity)<a href="https://agentmods.dev/skills/santoshkanthety/powerbi-agent/powerbi-fabric-capacity"><img src="https://agentmods.dev/badge/skills/santoshkanthety/powerbi-agent/powerbi-fabric-capacity/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/santoshkanthety/powerbi-agent/powerbi-fabric-capacity"><img src="https://agentmods.dev/badge/skills/santoshkanthety/powerbi-agent/powerbi-fabric-capacity.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 | $0.00135 | $0.02217 |
| Opus 5.5 | $0.00054 | $0.00887 |
| Sonnet 5.5 | $0.00027 | $0.00443 |
| Haiku 4.5 | $0.00014 | $0.00222 |
Grade A, and why
powerbi-fabric-capacity 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 16d 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 — 169 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Fabric capacity: CU accounting and cost-before-action
A Fabric capacity is a shared, metered pool. Every query, refresh, pipeline run, Spark job and Fabric App request draws capacity units (CU) from it, and a capacity pushed past its limit does not fail loudly — it throttles everyone on it, including reports nobody touched.
Two jobs:
- Attribute consumption — what is actually burning CU, per item and operation.
- Estimate before acting — say what an action will cost the capacity before running it.
Estimate before you act
Before any operation that draws meaningful CU on a shared or production capacity, tell the user the expected impact in one line, then wait for a yes. Treat these as triggering it:
- A full semantic model refresh, or any refresh of an import model over ~1 GB
- A Direct Lake model reframe or a large table reload
- A Spark job, notebook run or Livy session (see
powerbi-spark-livy) - A pipeline backfill, or any ETL over more than one partition
- Standing up a Fabric App on a shared capacity (see
powerbi-fabric-apps) - A bulk
faboperation across many items - A DAX query without a row bound against a large fact table
Shape of the statement, with real numbers where you have them:
Full refresh of
Sales Analytics(4.2 GB import, 18 partitions). Last comparable refresh consumed ~3,100 CU seconds — about 2.2 % of the F64 daily budget, running ~11 minutes. Current capacity utilisation is 71 %. Safe now; it would be marginal during the 08:00 business-hours peak. Proceed?
When you have no prior measurement, say that rather than inventing a figure: name the comparable operation you would measure, offer to measure it, and give the order of magnitude.
CU mechanics
| Concept | What it means in practice |
|---|---|
| CU second | The unit everything is billed in. A capacity's SKU number is its CU-per-second allowance; an F64 has 64 CU/s, so 5,529,600 CU seconds per day. |
| Smoothing | Consumption is spread forward rather than charged at the instant it happens — interactive operations over a short window, background operations (refreshes, pipelines, Spark) over 24 hours. This is why a heavy refresh shows up as a flat shelf across the next day, not a spike, and why yesterday's work can throttle you today. |
| Bursting | A single operation may consume faster than the SKU allows, finishing sooner; the excess is smoothed forward. Bursting makes operations fast and overage debt invisible until it throttles. |
| Overage / carry-forward | Smoothed future consumption that exceeds the allowance. The throttling stage is a function of how far forward the overage reaches, not of instantaneous load. |
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.
- 16d ago First seen · 169 lines · 0 tokens per session scan A 75a65148a353
powerbi-fabric-capacity is a skill published in the GitHub repository santoshkanthety/powerbi-agent (3 stars, last pushed 17d ago), licensed MIT. It adds 135 tokens to every session and 2,217 once invoked, about $0.0005 per session on Opus 5.5. 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 2026-10-07.
Other skills, from other repositories
seshat-bi
Route a BI project through Seshat BI's governed seven-stage readiness flow. Use when a user asks to inspect a retail source, initialize a Seshat project, find the truthful next action, validate readiness evidence, or stop at the correct human approval gate.
powerbi-workflows
Route guarded Power BI work -- design, native report authoring, semantic-model operations, published queries, QA, bounded formatting, and PBIP adoption -- to the correct Seshat or official Microsoft surface under Seshat BI's gates.
dbt-workflows
Route dbt intent to Seshat's governed shadow workflow or the official dbt Labs competence and execution owner, without bypassing readiness or evidence.
dagster-workflows
Route Dagster intent to Seshat's governed medallion workflow or the official Dagster competence owner without bypassing readiness, approvals, or evidence.
pbi-mcp-doctor
Use when a user asks whether or how to wire Microsoft's official Power BI MCP servers into a Seshat BI workspace: run the read-only environment doctor, map a task to the governed Power BI surface (including the official report-authoring skill), generate a safe read-only config template, or run the mocked read-only…
bi-analyst-knowledge
Use Seshat's public analyst-judgment knowledge to derive decision-questions, choose an analytical framing, and order a story from approved metric contracts and a committed source-profile.