中文

A Claude Code skill for looking up standard journal/magazine name abbreviations. Supports both ISO 4 and MEDLINE abbreviation standards, covering 25,000+ journals.

1. JabRef Database — 25,000+ journals, CC0 licensed, auto-downloaded on first run
2. AbbrevISO API — Algorithmic ISO 4 abbreviation based on LTWA (forward only)
3. NLM Catalog — Bidirectional lookup for biomedical journals (MEDLINE standard)

The skill bundle lives at skills/journal-abbrev/ (containing SKILL.md and jabbrv.py). The recommended path is the plugin marketplace — it handles updates for you:

# Claude Code plugin marketplace (recommended)
/plugin marketplace add Agents365-ai/365-skills
/plugin install journal-abbrev

# Any agent (Claude Code, Cursor, Copilot, …)
npx skills add Agents365-ai/365-skills -g

Also published on ClawHub and SkillsMP — each handles updates through its own marketplace.

Or install the skill bundle manually into any SKILL.md-aware platform:

Or just clone the repo and run the CLI directly — it is pure Python 3.9+ standard library with no third-party dependencies, and works on macOS, Linux, and Windows:

git clone https://github.com/Agents365-ai/journal-abbrev.git
cd journal-abbrev/skills/journal-abbrev
python3 jabbrv.py lookup "Nature Medicine"
python3 jabbrv.py schema              # full machine-readable CLI contract

Ask Claude directly:

• "What's the abbreviation for Nature Medicine?"
• "What journal is J. Biol. Chem.?"
• "Abbreviate all journal names in refs.bib"
• "Search for journals related to biolog chem"

# Auto-detect direction
python3 jabbrv.py lookup "Nature Medicine"

# Full name → abbreviation
python3 jabbrv.py abbrev "Journal of Biological Chemistry"

# Abbreviation → full name
python3 jabbrv.py expand "Nat. Med."

# Fuzzy search (paginated)
python3 jabbrv.py search "biolog chem" --limit 10 --offset 0

# Process BibTeX file (preview without writing first)
python3 jabbrv.py bib refs.bib --dry-run
python3 jabbrv.py bib refs.bib --output refs_final.bib
python3 jabbrv.py bib refs.bib --idempotency-key run-001   # safe to retry

# Batch lookup (one journal per line)
python3 jabbrv.py batch journals.txt
python3 jabbrv.py batch journals.txt --stream   # NDJSON, one result per line

# Cache management
python3 jabbrv.py cache status                # inspect local cache
python3 jabbrv.py cache update                # download missing files
python3 jabbrv.py cache update --dry-run      # preview what would be downloaded
python3 jabbrv.py cache rebuild --yes         # atomic destructive refresh
python3 jabbrv.py cache rebuild --dry-run     # preview what would be deleted, no writes

# Suppress stderr progress (envelope on stdout unchanged)
python3 jabbrv.py --quiet cache update

# Machine-readable CLI contract (for AI agents and automation)
python3 jabbrv.py schema
python3 jabbrv.py schema lookup

Stdout is a stable JSON envelope when not attached to a terminal (piped or captured by an agent runtime), and a human table/indent view when run on a TTY. Every response carries the same shape:

• Success: {"ok": true, "data": ..., "meta": {"schema_version", "cli_version", "cache", "latency_ms"}}
• Partial (batch): {"ok": "partial", "data": {"succeeded": [...], "failed": [...]}}
• Error: {"ok": false, "error": {"code", "message", "retryable", ...}}

The error.code field is stable. Notable codes:

Branch on error.code + error.retryable rather than exit code alone — exit 1 covers both transient and runtime errors. Force a format with --format json|table|human (or the back-compat --json); flags may appear before or after the subcommand. Full machine-readable listing: python3 jabbrv.py schema → data.error_codes.

Branch on error.code and error.retryable. Bound retries; back off on transient failures; do not retry definitive misses or validation errors. Reuse --idempotency-key on bib so a retried call returns the original envelope instead of re-running the rewrite:

import json, subprocess, time

def jabbrv_bib(path, key, *, max_attempts=4):
    for attempt in range(1, max_attempts + 1):
        p = subprocess.run(
            ["python3", "jabbrv.py", "bib", path,
             "--idempotency-key", key, "--format", "json"],
            capture_output=True, text=True,
        )
        env = json.loads(p.stdout)
        if env["ok"] is True:
            return env  # success (meta.idempotent_replay True on replays)
        err = env["error"]
        if not err["retryable"] or attempt == max_attempts:
            raise RuntimeError(f"{err['code']}: {err['message']}")
        time.sleep(2 ** attempt)  # exponential backoff

cache rebuild stages a fresh download into a sibling directory and only swaps it into place once every file succeeded. If any file fails, the existing cache is preserved and the response is error.code = upstream_unavailable (retryable). Add --yes to record explicit destructive intent in meta.confirmed; the CLI never prompts, so --yes is for audit policy, not for gating.

Set by the host/sandbox, not by agent argv — the trust-boundary half of agent-native design:

Schema introspection: python3 jabbrv.py schema → data.global_env. Each command's safety tier (read / write / destructive) is exposed as data.commands.<name>.mutates.

• Python 3.9+ (no third-party packages required)

• Discord: https://discord.gg/79JF5Atuk
• WeChat: scan the QR code below

If this skill helps you, consider supporting the author:

WeChat Pay

Alipay

Buy Me a Coffee

Give a Reward

Agents365-ai

• GitHub: https://github.com/Agents365-ai
• Bilibili: https://space.bilibili.com/441831884

MIT