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 agentmods add instructions/dginev/latexml-oxide/claude-mdgit clone --depth 1 https://github.com/dginev/latexml-oxideWrote 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/instructions/dginev/latexml-oxide/claude-md)<a href="https://agentmods.dev/instructions/dginev/latexml-oxide/claude-md"><img src="https://agentmods.dev/badge/instructions/dginev/latexml-oxide/claude-md.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 | $0.05369 | $0.05369 |
| Opus 5 | $0.02684 | $0.02684 |
| Sonnet 5 | $0.01074 | $0.01074 |
| Haiku 4.5 | $0.00537 | $0.00537 |
Grade A, and why
latexml-oxide CLAUDE.md 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 5d 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 — 298 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
This is a Perl-to-Rust translation project. Every translated entry must follow tightly the original semantics and nuances of the Perl source. Read the Perl source first, translate precisely; do not invent new abstractions, rename concepts, or simplify behavior unless documented as an intentional divergence in
docs/parity/OXIDIZED_DESIGN.md. The Perl code is the ground truth.
Active priorities: faithful parity + beyond-Perl arXiv runs
Two co-equal targets drive current work:
- Faithful translation of the original Perl LaTeXML. Strict parity at the
format/dump and package-loading boundary is maintained — the strict-
LoadFormatdump-parity mission is complete (zero-error inits, dumps match Perl; audit archived atdocs/archive/PERL_LOADFORMAT_AUDIT.md, the ~72-CS Perl-only long-tail residual tracked inSYNC_STATUS.md"Engine file open gaps"). Ongoing parity work is corpus-driven: mine fatal/error clusters from live runs, classify vs same-host Perl (canvas-triageskill), fix GENUINE-RUST-ONLY divergences faithfully. - Beyond-Perl improvement runs over arXiv. The production
cortex_workerfleet converts the full ~2.8M-doc arXiv corpus. Beyond-Perl levers: performance (docs/performance/PERFORMANCE.md/ARXIV_PERFORMANCE.md), reliability (STABILITY_WITNESSES.md), and the source-provenance showcase (issues #47/#92,SOURCE_PROVENANCE.md).
Test counts, gate status, and corpus throughput live in
docs/SYNC_STATUS.md, which is commit-tracked. Do not copy
those numbers here — the copy drifts.
Two traps when reading a test run:
- A fully green suite still prints
Error:lines to stderr. Several tests deliberately raise diagnostics to prove they get reported (thegraphics.rsworker-thread fold emitsfailed_to_convertfor a nonexistentw0.pdf; the Rhai script-binding tests emitboom). Judge a test run by itstest result:lines and exit code, never by grepping its output forError:— that heuristic belongs to conversion logs (below), and here it inverts. - The two
latexml_postvector-SVG tests self-skip silently and green unlessmutoolorpdftocairois on PATH, so a green local run does not by itself prove that branch ran. CI installs poppler/mupdf.
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.
- 5d ago First seen · 298 lines · 5,369 tokens per session scan A 20a838ef1a3a
latexml-oxide CLAUDE.md is an instructions file published in the GitHub repository dginev/latexml-oxide (11 stars, last pushed today), licensed CC0-1.0. It adds 5,369 tokens to every session, about $0.0268 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 instructions, from other repositories
tilawa AGENTS.md
Instructions for yazinsai/tilawa, covering tilawa, layout, sdk architecture, public surface (@tilawa/core) and build & test the sdk.
cli AGENTS.md
AGENTS.md instructions for googleworkspace/cli, covering agents.md, project overview, build & test, changesets and architecture.
ton-rs AGENTS.md
AGENTS.md instructions for ston-fi/ton-rs, covering ton-rs agent guide, repository, boundaries, public api and ton invariants.
codesage AGENTS.md
AGENTS.md instructions for iliaal/codesage, covering codesage, build, sanity check before pushing, crate map and search pipeline.
mcpls rust-config.instructions.md
Instructions for bug-ops/mcpls, covering type safety and idiomatic config and msrv and dependencies.
glory AGENTS.md
Instructions for glory-rs/glory, covering agents.md, what glory is, crate layout, feature flags and combinations and reactivity rules.