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/mathpluscode/bibtools/agents-mdgit clone --depth 1 https://github.com/mathpluscode/bibtoolsWhat 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 | $0.00706 | $0.00706 |
| Opus 5 | $0.00353 | $0.00353 |
| Sonnet 5 | $0.00141 | $0.00141 |
| Haiku 4.5 | $0.00071 | $0.00071 |
Grade A, and why
bibtools 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 yesterday.
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 — 51 lines — stays where its author put it; the contents beside it link to each section on GitHub.
bibtools
A bibliography toolkit for LaTeX, built as agent skills. Explore the repo to understand the project structure before making changes.
General principles
Bias toward action. Do not spend excessive time exploring files or asking clarifying questions when the task is clear. Start implementing and iterate.
Code changes
- When fixing bugs or implementing changes, do NOT fix things that aren't actually broken. If you're unsure whether something is a real bug, ask before implementing a fix. Never apply fixes for theoretical issues that can't occur in practice given the actual workflow.
- When making changes, always check README.md and other documentation files for references that need updating. Check for hardcoded magic numbers that should use existing config values.
Code review
When asked to do a code review, only flag real actionable issues. Do NOT over-engineer fixes or apply unnecessary changes. When in doubt, propose the fix and wait for user approval before implementing.
How it works
bibtidy
In Claude Code, invoke with /bibtidy refs.bib. In Codex, ask it to use the bibtidy skill on a .bib file. The agent reads the file, dispatches parallel subagents to verify entries against Google Scholar and CrossRef (bundled crossref.py), then applies fixes sequentially with targeted edits. Every change includes the original entry commented out and one or more source URLs for verification.
Versioning
Version is tracked in three files (.claude-plugin/marketplace.json, .claude-plugin/plugin.json, pyproject.toml). Update all three, then run uv lock to sync uv.lock.
Development
The skill's tools live under skills/bibtidy/tools/ in this repo. At runtime, determine which agent platform you are (Claude Code, Codex, etc.) and resolve the installed tools directory accordingly (e.g. ~/.claude/skills/bibtidy/tools for Claude Code, ~/.codex/skills/bibtidy/tools for Codex). To test local changes:
- Run
./tests/run_bibtidy_cc_tests.sh— this syncs local tools to~/.claude/skills/bibtidy/before invoking Claude - Run
./tests/run_bibtidy_codex_tests.shfor the Codex end-to-end path - Unit tests only:
uv run pytest tests/
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.
- yesterday First seen · 51 lines · 706 tokens per session scan A 2795db9c08d8
bibtools AGENTS.md is an instructions file published in the GitHub repository mathpluscode/bibtools (5 stars, last pushed 4mo ago), licensed MIT. It adds 706 tokens to every session, about $0.0035 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 2026-08-31.
Other instructions, from other repositories
academic-research-skills CLAUDE.md
Instructions for YuanyuanMa03/academic-research-skills, covering claude.md, project overview, repository structure, development commands and validation (run before committing).
scitex-python CLAUDE.md
Claude Code instructions for scitex-ai/scitex-python, covering workflow orchestration, 1. plan mode default, 2. subagent strategy, 3. self-improvement loop and 4. verification before done.
academic-tools-mcp CLAUDE.md
Instructions for hunter-heidenreich/academic-tools-mcp, covering claude.md, what this is, commands, code style & the format-on-edit hook and changelog & versioning.
devonthink-mcp CLAUDE.md
Claude Code instructions for TomBener/devonthink-mcp, covering copilot instructions, project structure, available tools, adding new tools and 1. create the tool file.
devonthink-mcp copilot-instructions.md
Copilot instructions for TomBener/devonthink-mcp, a project described as: DEVONthink MCP Server.
devonthink-mcp AGENTS.md
AGENTS.md instructions for TomBener/devonthink-mcp, a project described as: DEVONthink MCP Server.