← Back to Skills Marketplace
865
Downloads
4
Stars
0
Active Installs
9
Versions
Install in OpenClaw
/install ragflow-skill
Description
Use for RAGFlow dataset tasks: create, list, inspect, update, or delete datasets; upload, list, update, or delete documents; start or stop parsing; check par...
Usage Guidance
This skill appears to be a straightforward client for a RAGFlow HTTP API. Before installing: (1) Ensure RAGFLOW_API_URL points to a trusted RAGFlow instance; (2) limit the RAGFLOW_API_KEY scope/permissions and rotate it if possible; (3) be cautious when asking the agent to upload files — the scripts will read and transmit any local file you provide or reference via @path, so do not pass sensitive local files unless intended; (4) the SKILL.md requires explicit confirmation for deletes, but verify your operational workflow enforces that; (5) if you do not want the agent to call this skill autonomously, disable autonomous invocation for this skill in your agent settings.
Capability Analysis
Type: OpenClaw Skill
Name: ragflow-skill
Version: 1.0.8
The skill bundle provides a legitimate set of tools for interacting with the RAGFlow API to manage datasets, documents, and retrieval tasks. The scripts (e.g., `upload.py`, `search.py`, `datasets.py`) use standard Python libraries like `urllib` for network communication and include appropriate error handling and input validation. While some scripts allow reading local files via a `@path` syntax for configuration updates, this behavior is consistent with the stated purpose of the tool and does not show evidence of malicious intent or hidden data exfiltration.
Capability Assessment
Purpose & Capability
Name/description match the included scripts: dataset CRUD, document upload/update/delete, parsing control, status checks, search, and listing configured models. Declared requirements (python3, RAGFLOW_API_URL, RAGFLOW_API_KEY) are appropriate and proportionate for an HTTP API client to a RAGFlow service.
Instruction Scope
SKILL.md directs the agent to run only the bundled scripts and to prefer --json. The scripts operate against the declared RAGFLOW_API_URL and use the API key for Authorization. The scripts legitimately read files for uploads and accept @path JSON inputs for update operations — this is expected for upload/update behavior but means the skill can read and transmit any local file the agent is asked to upload or pass as @file. Guardrails require explicit confirmation for deletes, which is appropriate.
Install Mechanism
No install spec; this is an instruction+script bundle that requires python3 on PATH. Nothing is downloaded or written during install, lowering the installation risk.
Credentials
Only RAGFLOW_API_URL and RAGFLOW_API_KEY are required and primaryEnv is the API key — this is appropriate for communicating with a RAGFlow HTTP API. There are no unrelated secrets requested. (Note: some helper code like common.py may read runtime config from args/env or config files — the manifest references resolve_runtime_config/require_api_key but declared env requirements match the expected inputs.)
Persistence & Privilege
always:false and no install hooks are present. The skill does not request system-wide persistence or changes to other skills. The agent may invoke the skill autonomously (disable-model-invocation is false), which is the platform default; this is not combined with other red flags here.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install ragflow-skill - After installation, invoke the skill by name or use
/ragflow-skill - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.8
- Documentation has been streamlined for clarity and brevity.
- Usage instructions now focus on the main dataset, document, parsing, retrieval, and model listing operations.
- All output must follow the specifications in reference.md, with exact relay of returned fields.
- Delete actions explicitly require listing items for confirmation and only proceed with exact IDs.
- Updated sample command lines and clarified the use of --json and confirmation guardrails.
- Credentials and environment requirements are now stated more clearly.
v1.0.7
Version 1.0.7
- Updated SKILL.md for clarity, conciseness, and improved guardrails on delete and parse operations.
- Simplified output rules: always use `--json`, and relay all API fields exactly.
- Enforced output and workflow to strictly follow reference.md.
- Clarified skill activation triggers and typical usage examples.
- Added explicit requirements for credential configuration.
v1.0.6
ragflow-skill 1.0.6
- Rewrote skill usage section to focus on user intent and grouped use cases by dataset management, document operations, parsing control, knowledge retrieval, and model information
- Added strong requirements to comply with output formatting in reference.md (tables, status icons, templates, no fabricated progress)
- Clarified response handling for errors and output (must relay script/API fields exactly)
- Removed bulky trigger phrase lists in favor of succinct, language-agnostic intent mapping
- Highlighted that skill should activate based on action intent, not exact keyword or language
v1.0.5
- Removed the file `readme.md`.
- No other functional or workflow changes.
v1.0.4
- Added LICENSE and readme.md files to the repository.
- Updated documentation in SKILL.md to include instructions for repeated usage with --save-to-memory, --base-url, and --api-key-file options.
- Documented parse_status.py usage step in workflow and command examples.
- No changes to core logic; only documentation and license/readme additions.
v1.0.3
Version 1.0.3
- Rebranded skill from "ragflow-knowledge" to "ragflow-dataset-ingest" and updated naming conventions.
- Removed redundant or excessive procedural and security policy documentation.
- Added clear bilingual (English/Chinese) trigger phrase lists for each supported intent (dataset/document management, parsing, retrieval, model listing).
- Clarified and streamlined the supported workflow for invoking scripts and handling user requests.
- Removed the readme.md file.
v1.0.2
- Added a .gitignore file to the project.
- Updated SKILL.md: added an explicit env_requires section for required environment variables.
- No functional changes to code or workflow logic.
v1.0.1
Initial release with improved documentation, security notes, and output standards.
- New security and privacy section added: explains API key usage, data transmission, and environment configuration requirements.
- Renamed skill from `ragflow-dataset-ingest` to `ragflow-knowledge`.
- Added explicit instructions and examples for all key workflows: dataset management, document operations, parsing control, knowledge retrieval, and model listing.
- Reference to required output formatting added—including use of [reference.md] templates, tables with status icons, and error field handling.
- Lists all supported triggers/operations, when to use the skill, and necessary confirmation steps for delete actions.
- `.env.example` and extensive git metadata provided for initial setup guidance.
v1.0.0
Initial release of RAGFlow dataset and retrieval skill.
- Added scripts for managing datasets and documents: creation, listing, updating, deletion, and upload.
- Implemented document parsing control: start, stop, and status checking.
- Enabled search/retrieval of relevant chunks from datasets.
- Included support for listing and inspecting available LLM models.
- Supports both English and Chinese trigger phrases.
- Enforced confirmation steps for destructive actions (deletion) to prevent accidental data loss.
Metadata
Frequently Asked Questions
What is RAGFlow?
Use for RAGFlow dataset tasks: create, list, inspect, update, or delete datasets; upload, list, update, or delete documents; start or stop parsing; check par... It is an AI Agent Skill for Claude Code / OpenClaw, with 865 downloads so far.
How do I install RAGFlow?
Run "/install ragflow-skill" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is RAGFlow free?
Yes, RAGFlow is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does RAGFlow support?
RAGFlow is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created RAGFlow?
It is built and maintained by yingfeng (@yingfeng); the current version is v1.0.8.
More Skills