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 ychampion/cskill-agents --skill modality-preserving-binary-offloadgit clone --depth 1 https://github.com/ychampion/cskill-agentsWrote 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/ychampion/cskill-agents/modality-preserving-binary-offload)<a href="https://agentmods.dev/skills/ychampion/cskill-agents/modality-preserving-binary-offload"><img src="https://agentmods.dev/badge/skills/ychampion/cskill-agents/modality-preserving-binary-offload.svg" alt="Measured on agentmods" 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 | $0.00031 | $0.00396 |
| Opus 5 | $0.00015 | $0.00198 |
| Sonnet 5 | $0.00006 | $0.00079 |
| Haiku 4.5 | $0.00003 | $0.00040 |
Grade A, and why
modality-preserving-binary-offload 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 3d 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
SKILL: Modality-Preserving Binary Offload
Domain: tool-result-storage Trigger: Apply when a tool emits binary content that must be saved out of band without losing the file type or downstream usability. Source Pattern: Distilled from reviewed tool execution, streaming, persistence, and output-budget implementations.
Core Method
Preserve modality by deriving the output extension from MIME type, writing the raw bytes unchanged, and returning a saved-file message that names the type, size, and location. This ensures the resulting file can be opened by the correct downstream tool chain instead of being flattened into an opaque blob with the wrong extension. The offload path remains generic while still honoring modality-specific dispatch.
Key Rules
- Strip MIME parameters before extension lookup so content types like application pdf; charset binary still resolve correctly.
- Use conservative extension mapping: known types get their proper extension and unknown types fall back to
bin. - Write bytes as-is rather than stringifying them, or the saved artifact will no longer be usable by native readers.
- Return a human-readable saved-file message with MIME, size, and path so downstream agents know what was persisted.
Example Application
If an MCP server returns a PDF or spreadsheet, derive pdf or xlsx from the MIME type, persist the raw bytes into the tool-results directory, and emit a saved-file message so later steps can hand the file to a PDF reader or spreadsheet parser without guessing the format.
Anti-Patterns (What NOT to do)
- Do not save binary content with a generic text extension; downstream tooling dispatch often depends on the suffix.
- Do not stringify binary bytes before writing; that destroys the original modality and makes the file unusable.
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.
- 3d ago First seen · 28 lines · 31 tokens per session scan A c5850ca0203e
modality-preserving-binary-offload is a skill published in the GitHub repository ychampion/cskill-agents (36 stars, last pushed 5mo ago), licensed MIT. It adds 31 tokens to every session and 396 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 2026-09-03.
Other skills, from other repositories
document-skills-pdf
Extract and process PDF documents for data pipelines. Use when parsing research papers, extracting tables from reports, or converting PDFs to structured data.
hebrew-document-generator
Generate professional Hebrew documents including PDF, DOCX, and PPTX with full RTL support and proper Hebrew typography. Use when user asks to create Hebrew PDF, generate Israeli business documents, "lehafik heshbonit", "litstor hozeh", build Hebrew Word document, create Hebrew PowerPoint, or produce Israeli templates…
grape
Use Grape MCP for Codex context continuity in coding repositories. Use when a task needs repeated-turn context, omitted context restore, stale-context checks, invalidation checks, or safe continuity across branch and dirty-worktree changes.
pdftomarkdown
Converts PDF documents, Word files (.docx), CSV spreadsheets, and YouTube video URLs into clean, token-efficient Markdown with smart table and heading structure detection.
Use when tasks involve reading, creating, or reviewing PDF files where rendering and layout matter; prefer visual checks by rendering pages (Poppler) and use Python tools such as reportlab, pdfplumber, and pypdf for generation and extraction.
report-to-pdf
Narrow conversion skill. Invoke only when the user explicitly asks to convert an existing Data Analytics report, dashboard, or inline chart export into a PDF artifact.