← Back to Skills Marketplace
82
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install paper-anonymizer-academic-pdf-redaction
Description
Redact text from PDF documents for blind review anonymization
Usage Guidance
This is a coherent, instruction-only guide for redacting PDFs locally. Before using it: (1) Test on copies of documents — the verify function and thresholds are inconsistent (80% vs 70%) and the 1000-character check may be inappropriate for short papers. (2) Update pattern handling to use proper regular expressions when needed (PyMuPDF's page.search_for() expects literal text; wildcard patterns like '*@*.edu' won't work as written). (3) Be careful with the 'References' detection heuristic (it can misidentify pages). (4) Install PyMuPDF (fitz) from the official package index if you intend to run the code, and review dependency sources. There are no hidden network calls or credential requests in the instructions, but treat this as guidance (not packaged code) and review any code you run locally.
Capability Analysis
Type: OpenClaw Skill
Name: paper-anonymizer-academic-pdf-redaction
Version: 0.1.0
The skill bundle provides legitimate instructions and Python code for redacting academic papers using the PyMuPDF (fitz) library. The code includes safety checks to prevent over-redaction and verify document integrity, with no evidence of data exfiltration, malicious execution, or prompt injection attacks in SKILL.md.
Capability Assessment
Purpose & Capability
Name/description (redact PDFs for blind review) matches the instructions: they show how to find patterns in PDF text and apply redactions. No unrelated credentials, binaries, or config paths are requested.
Instruction Scope
Instructions remain narrowly scoped to opening, searching, redacting, and verifying PDFs locally. Issues: (1) The top-level rule asks to 'Check that 80%+ of original text remains' but the included verify_redaction() raises only if retention < 70% (inconsistent thresholds). (2) Pattern examples (e.g., '*@*.edu') are presented as wildcard patterns, but the sample PyMuPDF code uses page.search_for(pattern) which matches literal text, not shell-style wildcards—this could cause incorrect behavior if copy/pasted. (3) The heuristic for locating the References page (first page containing 'references') can misidentify sections. These are functional correctness concerns, not scope creep or exfiltration.
Install Mechanism
No install spec and no code files — the skill is instruction-only. It recommends using PyMuPDF but does not attempt to download or execute anything itself, minimizing install risk.
Credentials
The skill requests no environment variables, credentials, or config paths. All operations are local file I/O consistent with its purpose.
Persistence & Privilege
always:false and default invocation settings are used. The skill does not request persistent privileges or modify other skills or system configs.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install paper-anonymizer-academic-pdf-redaction - After installation, invoke the skill by name or use
/paper-anonymizer-academic-pdf-redaction - Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.0
Bulk publish from all-task-skills-dedup
Metadata
Frequently Asked Questions
What is academic-pdf-redaction?
Redact text from PDF documents for blind review anonymization. It is an AI Agent Skill for Claude Code / OpenClaw, with 82 downloads so far.
How do I install academic-pdf-redaction?
Run "/install paper-anonymizer-academic-pdf-redaction" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is academic-pdf-redaction free?
Yes, academic-pdf-redaction is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does academic-pdf-redaction support?
academic-pdf-redaction is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created academic-pdf-redaction?
It is built and maintained by lnj22 (@lnj22); the current version is v0.1.0.
More Skills