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 j4flmao/agent-skills --skill data-formatsgit clone --depth 1 https://github.com/j4flmao/agent-skillsWrote 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/j4flmao/agent-skills/data-formats)<a href="https://agentmods.dev/skills/j4flmao/agent-skills/data-formats"><img src="https://agentmods.dev/badge/skills/j4flmao/agent-skills/data-formats/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/j4flmao/agent-skills/data-formats"><img src="https://agentmods.dev/badge/skills/j4flmao/agent-skills/data-formats.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 | $0.00127 | $0.05020 |
| Opus 5 | $0.00063 | $0.02510 |
| Sonnet 5 | $0.00025 | $0.01004 |
| Haiku 4.5 | $0.00013 | $0.00502 |
Grade A, and why
data-formats 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 8d 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 — 510 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Data Formats
Purpose
Design efficient data storage and transfer using Apache Arrow, Parquet, Avro, and ORC formats with appropriate compression, schema evolution strategies, and high-performance transport via Arrow Flight.
Agent Protocol
Trigger
Exact user phrases: "Apache Arrow", "Parquet", "Avro", "ORC", "Arrow Flight", "columnar", "row-oriented", "compression", "schema evolution", "file format", "data format", "columnar storage", "data serialization", "row group", "arrow table", "IPC format", "Flight SQL".
Input Context
Before activating, verify:
- Data access patterns (full scan, column projection, row lookup, point queries)
- Storage target (S3, HDFS, local disk, memory, network transfer)
- Write patterns (append-heavy, overwrite partitions, streaming)
- Processing framework (Spark, DuckDB, pandas, Polars, Dremio, ClickHouse)
- Schema evolution requirements (add/drop/rename columns over time)
- Compression requirements (storage cost vs CPU cost)
Output Artifact
Data format specification with file layout, encoding, compression, and schema configuration as SQL, YAML, and Python.
Response Format
-- Parquet DDL with encoding and compression
-- Arrow table construction and IPC
-- File format configuration for Spark/DuckDB
No preamble. No postamble. No explanations. No filler/hedging/transitions. Compress output.
Completion Criteria
- Format selection justified by access pattern and workload
- Parquet row group size and page size configured
- Compression codec selected with rationale
- Schema evolution strategy documented
- Arrow Flight endpoint designed for transport
- Encoding selection for efficiency (dictionary, RLE, delta)
Max Response Length
4096
Workflow
Format Selection Guide
| Requirement | Columnar (Parquet, ORC) | Row-Oriented (Avro, JSON) | In-Memory (Arrow) |
|---|---|---|---|
| Access pattern | Column projection, full scan | Row-by-row, lookups | Analytical processing |
| Write pattern | Batch, append to partitions | Streaming, Kafka | In-process, batch |
| Compression ratio | High (column similarity) | Low-Medium | N/A (memory) |
| Schema evolution | Backward compatible | Full support | Requires copy |
| Zero-copy reads | No (file-based) | No | Yes |
| Inter-language | Any (file-based) | Any | C++, Python, R, Java, JS |
| Best for | Data lakes, analytics | Message queues, Kafka | Compute engines |
What ships with it
8 files 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.
- references/arrow-flight-protocol.md 9.5 KB
- references/columnar-format-guide.md 9.1 KB
- references/columnar-formats.md 9.0 KB
- references/compression-encoding.md 5.8 KB
- references/data-serialization-patterns.md 3.7 KB
- references/file-format-benchmarks.md 6.3 KB
- references/format-migration-strategies.md 4.6 KB
- references/schema-evolution.md 7.8 KB
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.
- 8d ago First seen · 510 lines · 127 tokens per session scan A ddb250d613d6
data-formats is a skill published in the GitHub repository j4flmao/agent-skills (23 stars, last pushed 6d ago), licensed MIT. It adds 127 tokens to every session and 5,020 once invoked, about $0.0006 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 2026-09-03.
Other skills, from other repositories
analytical-databases-expert
Design and query columnar analytical stores: DuckDB, ClickHouse and cloud warehouses, including file formats, partitioning, sort keys and cost control. Use when the user mentions DuckDB, ClickHouse, Parquet, columnar storage, OLAP, a data warehouse or lakehouse, analytical queries over large tables, or when the task…
arrow
Apache Arrow in-memory columnar format reference. Zero-copy data exchange, columnar memory layout with validity bitmaps, pyarrow Table/RecordBatch/compute, Arrow Flight RPC for high-performance transfer, Dataset API with predicate pushdown, pandas/DuckDB/Polars/Spark integration, Gandiva LLVM compiler, and ADBC…
duckdb
Use when analytical SQL must run in-process with no server: Parquet/CSV/JSON/Arrow queried in place, OLAP embedded in an app or notebook, a slow pandas groupby on multi-GB data, or S3/lakehouse data read without downloading. NOT a multi-user analytics server (that is clickhouse-analytics), NOT an app's transactional…
data-pipeline
ETL and data transformation: CSV, JSON, Parquet, NDJSON processing with filtering, mapping, joining, aggregating.
arrowspace
Spectral vector search using graph Laplacian eigenstructure. Use when cosine/L2 similarity misses latent structure in your embeddings.
data-report
Turns CSV, Excel, or JSON data into a polished visual report page.