hfpclawer Citation Audit

Part of the Exo suite — literature (hfpclawer) → experiments (expflow-pde) → proofs (omega-architect). Three independent CLIs that meet through files and CLI calls, never imports. Entry skill: exo-suite-linkage (wiring, cost tiers low → medium → high, degradation ladder). Install: uv tool install hfpclawer · uv tool install expflow-pde · uv tool install "omega-architect @ git+https://github.com/diamond2nv/omega-architect@v0.2.3"

🔒 Sanitization: This skill ships in the public repo. Never embed private LAN IPs, real person names, or machine codenames in examples — use <placeholder> / Jane Doe / dev@example.com. Real values live in the repo's gitignored .hermes/internal-guide.md. See repo AGENTS.md "Public-Release Sanitization".

Verify whether a cited academic paper actually exists, using a three-tier pipeline that degrades gracefully when local data or remote APIs are unavailable.

Who this is for: Researchers, reviewers, and literature-survey authors who need to check whether a citation refers to a real paper.

Overview

The audit engine tries three sources in order, stopping at the first confirmation:

                    ┌──────────────────────────┐
 User:              │  hfpclawer audit verify   │
 "Is this paper     │  "Fourier Neural Operator"│
 real?"             └─────────────┬────────────┘
                                  │
                    ┌─────────────┼─────────────┐
                    ▼             ▼              ▼
              ┌─────────┐  ┌──────────┐  ┌──────────┐
              │ L1:     │  │ L2:      │  │ L3:      │
              │ Local   │→ │ Semantic │→ │ OpenAlex │
              │ FTS5 DB │  │ Scholar  │  │          │
              │ (1ms)   │  │ (200ms)  │  │ (200ms)  │
              └─────────┘  └──────────┘  └──────────┘

Each source independently reports one of four statuses:

• VERIFIED — the paper exists in this source
• SUSPECTED — possible match (similar title, but not exact)
• NOT_FOUND — no match found
• ERROR — source unavailable (no local DB / API rate-limited)

When to Use

• A user cites a paper you cannot find — verify its existence
• You team is writing a survey / literature review — batch audit the reference list
• You downloaded an LLM-generated paper and need to fact-check its citations
• You want to know whether a paper is a known arXiv preprint or a non-existent hallucination

Prerequisites

Python ≥ 3.10. Pick the install that matches how you work — uv is recommended, because the CLI then lives in its own environment (no conflicts with your project's dependencies):

# 1) Recommended — uv tool: isolated CLI install, `hfpclawer` on your PATH
uv tool install hfpclawer
hfpclawer init                    # writes config.yaml

# 2) Try it without installing anything (ephemeral, one-off runs)
# pin the version — `uvx`/`uv tool run` reuse an installed tool env (may run an older
# release), and an unpinned launch is a supply-chain (rug-pull) risk
uvx "hfpclawer==0.19.0" --help

# 3) Inside an existing project / venv (uv-managed)
uv pip install hfpclawer

# 4) No uv yet — pip and pipx both work
pip install hfpclawer             # or: python -m pip install hfpclawer
pipx install hfpclawer            # CLI-style install, functionally like `uv tool`

Optional extras — the core install stays deliberately small:

⚠️ nlp extra + PyPI (checked against the published 0.19.0 metadata): PyPI strips the direct-URL spaCy model, so that extra installs spacy only — fetch the model yourself with python -m spacy download en_core_web_sm. The loader falls back configured → en_core_web_md → en_core_web_sm and logs one actionable hint when none is present: entity enrichment degrades, nothing else breaks.

Where to find it: repo https://github.com/diamond2nv/hfpapers-crawler · PyPI https://pypi.org/project/hfpclawer/ · registry: clawhub inspect <slug>

• No API keys needed for basic use (S2 + OpenAlex use anonymous tier)
• Optional: Set S2_API_KEY env var for 10x faster Semantic Scholar lookups
• Optional: Set OPENALEX_POLITE_EMAIL env var for 10x faster OpenAlex lookups
• Optional: Clone arxiv-metadata-service repo for L1 local FTS5 (see references/local-db-setup.md)

Quick Start

1. Verify a Single Citation (most common)

# Auto mode: tries local DB first, then Semantic Scholar, then OpenAlex
hfpclawer audit verify "Fourier Neural Operator for Parametric Partial Differential Equations"

# Short title works too — includes substring fallback
hfpclawer audit verify "Fourier Neural Operator"

# Exact arXiv ID
hfpclawer audit verify --arxiv-id 2010.08895

2. Use a Specific Source

# Local FTS5 only (needs arxiv_meta.db)
hfpclawer audit verify "Attention Is All You Need" --source local

# Semantic Scholar only
hfpclawer audit verify "Attention Is All You Need" --source s2

# OpenAlex only
hfpclawer audit verify "Attention Is All You Need" --source openalex

3. Check a Reference List from File

# Save citations in a text file, one per paragraph
cat > refs.txt << 'EOF'
The FNO paper (arXiv:2010.08895) shows promising results.
PINNs were introduced by Raissi et al. (2019) "Physics-informed neural networks".
EOF

hfpclawer audit --refs refs.txt

Output Format

Each result shows:

• [OK] VERIFIED — paper confirmed; includes title, authors, source
• [?] SUSPECTED — possible but uncertain; shows top matches
• [NF] NOT_FOUND — no evidence of this paper
• [ERR] ERROR — source unavailable (DB not found, rate limited)

[OK] VERIFIED
  Title: Fourier Neural Operator
  Authors: Zongyi Li, Nikola Kovachki, Kamyar Azizzadenesheli, ...
  Sources: openalex: VERIFIED

How Statuses Are Determined

Title matching: Title similarity uses difflib.SequenceMatcher on normalized (lowercase, punctuation-stripped) titles. Short titles that are substrings of longer titles also pass the 0.70 threshold.

Batch Modes

From a Text File

hfpclawer audit --refs references.txt

The parser detects:

• arXiv:XXXX.XXXXX identifiers
• "Title" (Author, Year) patterns
• Author (Year) "Title" patterns

Via Python API

from hfpclawer.citation_audit import check_citation

result = check_citation(
    "Fourier Neural Operator",
    authors_hint="Li",
    year_hint=2020,
    source="auto",       # or "local" / "s2" / "openalex"
)
print(result["status"])  # VERIFIED | NOT_FOUND | ERROR
print(result.get("authors", "N/A"))
print(result.get("per_source", {}))  # Per-source breakdown

Hermes Agent Environment

hfpclawer is built to run inside Hermes Agent (and OpenCode) as a first-class tool — the agent discovers the skill, and every command below is callable without leaving the session.

1. Install the skill — place this folder under ~/.hermes/skills/research/<slug>/ (or install from ClawHub: clawhub inspect <slug> --file SKILL.md); Hermes loads it automatically and skill_view(name='<slug>') returns this file.

2. Register the MCP server so the agent calls the CLI as tools:

# ~/.hermes/config.yaml
mcp:
  servers:
    hfpclawer:
      command: "hfpclawer"
      args: ["mcp"]        # stdio mode — Hermes native MCP client

For OpenCode / debugging use HTTP mode: hfpclawer mcp --mode http --port 8765.

3. Environment variables — every one is optional; the pipeline runs with none set:

4. Cost profile — the mechanical layer (search / dedup / verify / audit / Zotero) needs low-cost by design — not zero-cost. The core path (search / dedup / verify / audit / Zotero) issues no LLM call and needs no API key, so the marginal cost per run is small; it is still not zero — bandwidth, disk and CPU are spent, and rate-limited upstreams (arXiv / OpenAlex / Semantic Scholar) can throttle or expect a key at volume. LLM features (sniff, abstract triage) are opt-in and metered where they run; rank trains locally with lightgbm. Describe the cost as low, and keep the LLM steps explicit — do not advertise the stack as zero-token.

5. Keep private data private — real author lists, ORCIDs and Zotero ids belong in config.local.yaml (gitignored) or ~/.hfpclawer/profile.yaml, never in tracked files. The repo-side profile (the hfpclawer: block in a project AGENTS.md) is deliberately public-safe: neutral academic keywords only.

Common Pitfalls

1. Short/two-word queries may fail L1 because FTS5's porter stemmer requires actual content words. Use at least 3-4 significant words for local DB queries.
2. Semantic Scholar rate-limits aggressively without API key (~1 req/s, ~100 req/day anonymous). Set S2_API_KEY for production use.
3. OpenAlex polite pool is free and gives 10 req/s — set OPENALEX_POLITE_EMAIL to your institution email.
4. No L1 without arxiv-metadata-service: The local FTS5 DB requires git clone of the separate arxiv-metadata-service repo. Without it, the auto chain starts at L2 (slower but still works).

Verification Checklist

• Single citation works: hfpclawer audit verify "Known Paper Title"
• arXiv ID works: hfpclawer audit verify --arxiv-id 2010.08895
• Non-existent paper returns NOT_FOUND
• Network errors return ERROR (not crash)
• Batch mode processes multiple citations from file
• hfpclawer audit verify --help shows source options