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/pachterlab/paperpush/agents-mdgit clone --depth 1 https://github.com/pachterlab/paperpushWhat 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 | $0.02328 | $0.02328 |
| Opus 5 | $0.01164 | $0.01164 |
| Sonnet 5 | $0.00466 | $0.00466 |
| Haiku 4.5 | $0.00233 | $0.00233 |
Grade A, and why
paperpush AGENTS.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 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.
How it starts
The opening of the file, as written. The whole thing — 190 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Guidance for any AI agent (Codex, Claude, or otherwise) that decides to use
paperpush to submit a manuscript to a venue - a journal, preprint server, or conference -
on a user's behalf.
The one rule: you are the extractor
paperpush splits submission into two layers:
- You (the agent) propose values. You read the manuscript files and decide what each submission field should contain.
- The
paperpushCLI writes and validates. A deterministic core does all the writing, policy gating, path resolution, and validation, so you cannot overwrite a policy field or produce an invalid file.
Do your own autofilling. Read the manuscript yourself and hand the CLI a
values.json. Do not reach for paperpush autofill --engine api — that
path calls the Anthropic API and needs an ANTHROPIC_API_KEY. You already have
the manuscript in front of you; extract from it directly. Use --engine api
only if the user has explicitly set an API key and asked for it.
Before anything else: check the user is logged in
The pipeline ends in a real browser sign-in, and that sign-in is the one step you cannot do for the user — it may need their password, ORCID, a CAPTCHA, or a two-factor code. Check it first, before you read a single manuscript file, so the user can log in while you work instead of discovering the blocker at the end:
paperpush login --list
It prints Not logged in to any venues. or a Logged in to: block with one
<venue>: <who> line per stored credential. If the target venue is not in that
list, stop and tell the user right away — in your very next message — to run the
login themselves in their terminal:
paperpush login <venue>
Say plainly that this opens a browser and is theirs to complete, and that you
will keep going in the meantime. Do not run paperpush login yourself, and do
not treat a missing login as a reason to skip the extraction work — do the
.sub creation, autofill, and validation regardless, then re-check
paperpush login --list before the submit step.
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 · 190 lines · 2,328 tokens per session scan A c5ce57ac2962
paperpush AGENTS.md is an instructions file published in the GitHub repository pachterlab/paperpush (161 stars, last pushed 6d ago), licensed BSD-2-Clause. It adds 2,328 tokens to every session, about $0.0116 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
mcp-pandoc AGENTS.md
Instructions for vivekVells/mcp-pandoc, covering agents.md, what this is, where things are, deeper context and commands.
excel-mcp-server CLAUDE.md
Instructions for TangentDomain/excel-mcp-server, covering excel mcp server, 技术栈, 目录结构, mcp 工具 → python api 映射 and sql 核心原则.
academic-ref-inserter CLAUDE.md
Instructions for fuwenhao12/academic-ref-inserter, covering academic reference inserter (academic-ref-inserter), quick start, full pipeline, or step by step and supported formats.
agentic-resume-builder CLAUDE.md
Claude Code instructions for WtotdeD/agentic-resume-builder, covering agentic-resume-builder development guidelines, what this repo is, active technologies, package manager and validation.
fcp-sheets CLAUDE.md
Instructions for os-tack/fcp-sheets, covering fcp-sheets, project overview, architecture, key patterns and commands.
Doc-Mcp CLAUDE.md
Instructions for mosif16/Doc-Mcp, covering claude.md, project overview, build commands, build (requires rust 1.76+) and run tests.