Skills scanlume-ocr-api
📦
scanlume-ocr-api
Content revision r1 High Risk 🌐 Network access🔑 Env variables⚙️ External commands
Extract Text with Scanlume OCR
Teams often need reliable OCR for screenshots, receipts, and image-based tables. This skill guides Claude, Codex, and Claude Code through Scanlume API calls and structured OCR outputs.
Supports: Claude Codex Code(CC)
Stop for confirmation before installing.
Review the plan and obtain explicit user consent before changing files.
Install with my Agent
Copy this request to your Agent. It includes the canonical Skill page and manifest.
Agent request
Review the Skillstore skill "scanlume-ocr-api" from https://skillstore.io/skills/daanaagua-scanlume-ocr-api.md and its manifest at https://skillstore.io/api/skills/daanaagua-scanlume-ocr-api/manifest. Verify the artifact. Stop and obtain explicit user consent before installing or changing files.Your Agent should still show its plan and request any confirmation required by the security policy.
Agent-readable resources
Use these links when an AI agent, crawler, or script needs clean context instead of reading the full page.
Test it
Using "scanlume-ocr-api". A screenshot of a receipt with item names, prices, and totals.
Expected outcome:
A clean text extraction with line items, subtotal, tax, total, and unclear fields called out.
Using "scanlume-ocr-api". An image of a small sales table with grouped rows.
Expected outcome:
A structured table summary with columns, row groups, record counts, and any notes from the formatted OCR response.
Using "scanlume-ocr-api". A product settings screenshot that contains headings and paragraphs.
Expected outcome:
A Markdown-ready document that preserves headings, paragraphs, and readable order.
Security Audit
High Riskv4 • 7/5/2026 Open versioned report
Review found a legitimate OCR integration that sends local image data to the Scanlume API and handles a bearer API key. Most Markdown backtick findings are false positives, but the helper has real network, secret-handling, and command-execution risk. The main added concern is that an overridden base URL can receive the Authorization header.
5
Files scanned
667
Lines analyzed
23
Review items
0
False positives ignored
Confirmed security concerns (2)
High
API Key Exposure Through Base URL Override
The helper accepts an explicit or environment-provided base URL and sends Authorization bearer credentials to that computed endpoint. A malicious invocation could redirect the Scanlume API key to an untrusted server.
The code directly combines base URL override support with bearer-token request construction, which is a clear credential redirection risk.
Medium
Local Image Content Uploaded To External OCR Service
The helper reads a local file, encodes it as a data URL, and posts it to the Scanlume OCR API. Sensitive screenshots or documents may be disclosed if user consent and data handling are unclear.
The helper reads local bytes, builds a base64 data URL, and submits the payload through the API request path.
Capability review items (23)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
High
Generic API/secret keys
value = explicit or os.environ.get("SCANLUME_API_KEY")
The helper reads a Scanlume API key from the environment. This is legitimate for authentication, but it is real secret access.
High
Generic API/secret keys
def perform_request(base_url: str, api_key: str, payload: dict[str, str], timeout: int) -> dict[str,
The request helper accepts an api_key parameter that is later used for authentication. This is expected, but it is part of the credential transmission path.
High
Generic API/secret keys
"Authorization": f"Bearer {api_key}",
The Authorization header is built from the API key and sent with the request. If the endpoint is overridden, the credential can be exposed.
High
Generic API/secret keys
parser.add_argument("--api-key", help="Override SCANLUME_API_KEY.")
The CLI accepts an API key argument. Command-line secrets can leak through shell history, logs, or process listings.
High
Generic API/secret keys
api_key = require_api_key(args.api_key)
The main flow loads the API key before making the OCR request. This is legitimate, but it confirms runtime secret handling.
High
Generic API/secret keys
response_json = perform_request(base_url, api_key, payload, args.timeout)
The loaded API key is passed into the outbound request. This is expected for Scanlume auth, but it is credential transmission over the network.
Medium
Ruby/shell backtick execution
5. If the user gives a local file path, prefer `scripts/scanlume_ocr.py` to build the data URL and c
The workflow tells the agent to prefer the local Python helper when given a local file path. This can read local files and submit their contents to an external API.
Medium
Ruby/shell backtick execution
- Use `python scripts/scanlume_ocr.py <path> --mode formatted --output md` for a local table image.
The helper command example runs a Python script on a user-provided path. This is intended, but it is real external command execution.
Medium
Ruby/shell backtick execution
- Use `python scripts/scanlume_ocr.py <path> --mode simple --output txt` for plain text extraction.
The helper command example runs a Python script on a user-provided path. This is intended, but it is real external command execution.
Medium
Ruby/shell backtick execution
5. Se o usuario fornecer um caminho local, prefira `scripts/scanlume_ocr.py` para gerar a data URL e
The Portuguese workflow tells the agent to prefer the local Python helper for user-provided paths. This can read local files and call the API.
Medium
Ruby/shell backtick execution
- Use `python scripts/scanlume_ocr.py <path> --mode formatted --output md` para uma imagem local com
The Portuguese helper example runs the local Python script for formatted OCR. This is real external command execution.
Medium
Ruby/shell backtick execution
- Use `python scripts/scanlume_ocr.py <path> --mode simple --output txt` para extracao simples de te
The Portuguese helper example runs the local Python script for simple OCR. This is real external command execution.
Low
Hardcoded URL
short_description: "Image OCR / OCR de imagem for https://www.scanlume.com/."
The agent metadata names the external Scanlume service. This is expected for the skill, but it signals third-party OCR processing.
Low
Hardcoded URL
default_prompt: "Use $scanlume-ocr-api to call https://www.scanlume.com/ for image OCR and return th
The default prompt directs the agent to call Scanlume for OCR. User image data can leave the local environment during normal use.
Low
HTTP client library
def perform_request(base_url: str, api_key: str, payload: dict[str, str], timeout: int) -> dict[str,
The helper defines an outbound HTTP request function for the OCR endpoint. This is intended behavior, but it creates external data transfer risk.
Low
HTTP client library
response_json = perform_request(base_url, api_key, payload, args.timeout)
The helper invokes the outbound request after building the OCR payload. Local image content is submitted to a network API.
Low
Hardcoded URL
DEFAULT_BASE_URL = "https://api.scanlume.com"
The script hardcodes the default Scanlume API host. This is an expected external dependency, but it confirms network egress.
Low
Hardcoded URL
parser.add_argument("--base-url", help="Override SCANLUME_BASE_URL. Default / Padrao: https://api.sc
The argument help documents a base URL override that controls the outbound endpoint. Misuse can redirect OCR payloads away from the official API.
Low
Python environment access
value = explicit or os.environ.get("SCANLUME_BASE_URL") or DEFAULT_BASE_URL
The script reads SCANLUME_BASE_URL from the environment to decide the outbound endpoint. That configuration can affect where data is sent.
Low
Python environment access
value = explicit or os.environ.get("SCANLUME_API_KEY")
The script reads SCANLUME_API_KEY from the environment. This is expected API authentication, but it handles a sensitive credential.
Low
Hardcoded URL
description: Use when calling the Scanlume OCR API for screenshots, JPG, PNG, or image-based tables,
The skill description identifies Scanlume as the external OCR provider. Normal use can send user images to that third party.
Low
Hardcoded URL
Use this skill when the task is specifically about calling the public OCR API behind [https://www.sc
The instructions say the skill is for calling the public Scanlume OCR API. This confirms intended external network use.
Low
Hardcoded URL
Use este skill quando a tarefa for especificamente chamar a API publica de OCR do [https://www.scanl
The Portuguese instructions also direct use of the public Scanlume OCR API. This confirms intended external network use.
Risk Factors
🌐 Network access (12)
🔑 Env variables (15)
references/api-contract.md:17 references/api-contract.md:97 scripts/scanlume_ocr.py:21 scripts/scanlume_ocr.py:26 scripts/scanlume_ocr.py:25 scripts/scanlume_ocr.py:26 scripts/scanlume_ocr.py:29 scripts/scanlume_ocr.py:30 scripts/scanlume_ocr.py:70 scripts/scanlume_ocr.py:77 scripts/scanlume_ocr.py:152 scripts/scanlume_ocr.py:167 scripts/scanlume_ocr.py:174 SKILL.md:27 SKILL.md:78
⚙️ External commands (38)
SKILL.md:17 SKILL.md:18 SKILL.md:19 SKILL.md:20 SKILL.md:21 SKILL.md:26 SKILL.md:27 SKILL.md:28 SKILL.md:29 SKILL.md:30 SKILL.md:37 SKILL.md:42 SKILL.md:46 SKILL.md:50 SKILL.md:51 SKILL.md:52 SKILL.md:53 SKILL.md:59 SKILL.md:60 SKILL.md:68 SKILL.md:69 SKILL.md:70 SKILL.md:71 SKILL.md:72 SKILL.md:77 SKILL.md:78 SKILL.md:79 SKILL.md:80 SKILL.md:81 SKILL.md:88 SKILL.md:93 SKILL.md:97 SKILL.md:101 SKILL.md:102 SKILL.md:103 SKILL.md:104 SKILL.md:110 SKILL.md:111
Audited by: codex View Audit History →
Share & cite this report
Share the versioned assessment report, neutral badge, embed card, and citations. Skillstore reports evidence without deciding whether this Skill is safe.
Copy report link
https://skillstore.io/skills/daanaagua-scanlume-ocr-api/audits/4?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdown badge
[](https://skillstore.io/skills/daanaagua-scanlume-ocr-api?utm_source=security_passport_badge)HTML badge
<a href="https://skillstore.io/skills/daanaagua-scanlume-ocr-api?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/daanaagua-scanlume-ocr-api/security.svg" alt="Skillstore security assessment" loading="lazy"></a>Embed card
<iframe src="https://skillstore.io/embed/skills/daanaagua-scanlume-ocr-api.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>Academic citations (APA · BibTeX · CFF)
APA citation
daanaagua. (2026). scanlume-ocr-api security audit report (audit version 4) [Author version unspecified]. Skillstore. https://skillstore.io/skills/daanaagua-scanlume-ocr-api/audits/4BibTeX citation
@techreport{daanaagua-daanaagua-scanlume-ocr-api-2026,
author = {daanaagua},
title = {scanlume-ocr-api security audit report (audit version 4)},
institution = {Skillstore},
year = {2026},
number = {4},
url = {https://skillstore.io/skills/daanaagua-scanlume-ocr-api/audits/4},
note = {Author version unspecified}
}CITATION.cff
cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "scanlume-ocr-api security audit report (audit version 4)"
version: "unspecified"
type: report
authors:
- name: "daanaagua"
date-released: "2026-07-05"
url: "https://skillstore.io/skills/daanaagua-scanlume-ocr-api/audits/4"
identifiers:
- type: other
value: "skillstore:daanaagua-scanlume-ocr-api:audit:4"
description: "Skillstore immutable audit report identifier"
Skillstore Score
Why this score Evidence Confidence: Medium 64
Architecture
85
Maintainability
87
Content
69
Community
83
Spec Compliance
What You Can Build
Convert Screenshots to Text
Extract plain text from interface screenshots for notes, support tickets, or QA records.
Recover Tables from Images
Use formatted OCR to preserve table structure from receipts, reports, or captured spreadsheets.
Automate OCR API Calls
Prepare data URLs, call the Scanlume endpoint, and return the best response field for downstream work.
Try These Prompts
Extract Plain Text
Use Scanlume OCR on this local image and return only the extracted plain text. Use simple mode unless structure is needed.
Preserve Table Structure
Use formatted Scanlume OCR on this table image. Return a clean table summary and note any uncertain cells.
Choose the Best Output
Inspect the image goal, choose simple or formatted OCR, and explain the selected mode before returning the final output.
Build an OCR Workflow
Create a repeatable Scanlume OCR workflow for a folder of screenshots. Include mode choice, output field selection, and error handling guidance.
Best Practices
- Use formatted mode when layout, headings, or tables matter.
- Use environment-based secrets instead of passing API keys directly in prompts.
- Confirm user consent before sending sensitive images to the external OCR API.
Avoid
- Do not claim public PDF OCR API support for this helper.
- Do not send private screenshots unless the user accepts external processing.
- Do not override the API base URL unless the endpoint is trusted.
Frequently Asked Questions
What image types does this skill target?
It targets local screenshots, JPG files, PNG files, and other image files with an inferred MIME type.
Does it support PDF OCR?
No. The helper blocks PDFs because the public Scanlume PDF API is not generally available.
Which tools can use this skill?
The generated metadata lists Claude, Codex, and Claude Code as supported tools.
What is the difference between simple and formatted mode?
Simple mode returns raw text. Formatted mode preserves structure, Markdown, HTML, blocks, and table summary fields.
Does the skill need an API key?
Yes. The helper requires a Scanlume API key through the environment or an explicit argument.
Can it OCR remote file URLs?
No. The documented public API expects a full base64 data URL, not a remote file URL.