← Back to Skills Marketplace
216
Downloads
2
Stars
0
Active Installs
10
Versions
Install in OpenClaw
/install oasis-audio
Description
Oasis Audio is an AI audio narration generator that transforms your current state of mind, content you want to digest, or recent life events into a personali...
Usage Guidance
What to know before installing: this skill will read your local agent session files and daily-memory/USER.md files (~/.qclaw and ~/.openclaw) to tailor audio output — that behavior is declared in the README and implemented in context_collector.py. The composed prompt (not the raw files) is sent to an external service at eagle-api.xplai.ai; the skill implements heuristics to redact common secrets and will pause for explicit confirmation if it detects sensitive content. It also stores a local consent file (.oasis_audio_consent.json) and an audit.log that contains the prompt text (after redaction). Consider: (1) whether you are comfortable allowing a third-party API to receive selected, composed text derived from your local conversations; (2) whether you trust xplai.ai as the destination; (3) using the provided --dry-run preview and refusing to persist consent if you want tighter privacy; and (4) inspecting/clearing the audit.log and consent file if you later revoke consent. If you need stronger guarantees (no data leaving your machine), do not grant the skill consent to read local session files or do not use it.
Capability Analysis
Type: OpenClaw Skill
Name: oasis-audio
Version: 1.0.9
The skill generates personalized audio by accessing sensitive local data, including full conversation histories and user profiles located in `~/.openclaw` and `~/.qclaw` via `context_collector.py`. While the implementation includes significant privacy safeguards—such as mandatory user consent, local-only processing of raw data, and automated regex-based redaction of credentials (SSH keys, API tokens) and PII in `xplai_gen_audio.py`—the inherent capability to ingest and summarize private user interactions for transmission to an external endpoint (https://eagle-api.xplai.ai) is a high-risk behavior. The code appears professionally written without clear malicious intent, but the broad access to personal history warrants a cautious classification.
Capability Tags
Capability Assessment
Purpose & Capability
The skill claims to produce personalized audio and therefore legitimately needs to read local conversation sessions, daily memories, and a USER.md to derive personalization anchors. The code (context_collector.py) explicitly targets ~/.qclaw and ~/.openclaw session/memory files and extracts structured profile fields — this matches the description.
Instruction Scope
SKILL.md and the code instruct the agent to read local session files and memory to build personalization context and then to compose a single outgoing prompt which is sent to https://eagle-api.xplai.ai. The skill includes a sensitive-content preview and an opt-in consent flow before sending if heuristics find sensitive content. This scope is consistent with personalization but is privacy-sensitive: local conversation data will be parsed and portions may be incorporated into the outgoing prompt unless the sensitive-content checks stop it.
Install Mechanism
This is an instruction+script-only skill with no install spec and no downloads; it invokes Python scripts already included. No external installers or third‑party package downloads are requested, which is low-risk for install-time behavior.
Credentials
The skill requests no environment variables or external credentials (API calls appear to be unauthenticated). It does access local files (~/.qclaw and ~/.openclaw and workspace files) which is proportionate to the stated goal of personalization. It also writes two local files (.oasis_audio_consent.json and audit.log) to persist consent and an audit trail.
Persistence & Privilege
The skill does not demand always:true or elevated platform privileges. It does persist a consent record (.oasis_audio_consent.json) and an audit.log containing the composed prompt (post-redaction). Persisting prompts and consent locally is reasonable for this functionality, but this storage can retain potentially sensitive derived content if the user authorizes sending sensitive prompts.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install oasis-audio - After installation, invoke the skill by name or use
/oasis-audio - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.9
**Oasis Audio 1.0.9 Changelog**
- Added sensitive information preview and explicit user confirmation before transmitting any potentially sensitive prompt.
- Introduced a one-time, first-use consent notice that must be acknowledged by the user before their local data can be used for personalized audio.
- Clarified description and execution policy around local-only context processing, privacy, and confirmation requirements.
- Updated context source detection (only `~/.qclaw/` and `~/.openclaw/` are checked by default).
- Refined documentation to streamline user instructions, prioritizing user consent and sensitive-content handling.
v1.0.8
Oasis Audio 1.0.8
- Added `audio_modes.md` detailing all 9 reference audio modes, including triggers and content suggestions.
- Added `text_architecture.md` describing the 7-layer framework for structuring and composing audio prompts.
- Expanded and clarified usage, execution, and personalization policies in documentation.
- Added required config path metadata for improved auto-detection in OpenClaw, EasyClaw, and OpenClaw sessions.
- Improved, more concise guidance on user need inference and context collection workflow.
v1.0.7
Oasis Audio 1.0.7
- No functional or documentation changes in this release.
- All logic, usage, and setup remain unchanged from the previous version.
v1.0.6
## oasis-audio 1.0.6 changelog
- Adds clarification that all conversation analysis runs locally; only a ~1000-character composed summary is sent to the external API.
- Specifies which local session directories are read for personalization (`~/.qclaw/`, `~/.easyclaw/`, `~/.openclaw/`).
- Details the initial 210-second wait after audio generation is triggered, and limits status polling to 10 times (up to ~12 minutes in total).
- Confirms that raw conversation data and USER.md are never transmitted outside the device.
v1.0.5
- Updated to version 1.0.4
- Expanded and clarified the skill description for better readability.
- Minor reformatting and renaming of the instruction header to "Oasis Audio".
- No functional changes or file modifications detected.
v1.0.4
- Updated to version 1.0.4
- Expanded and clarified the skill description for better readability.
- Minor reformatting and renaming of the instruction header to "Oasis Audio".
- No functional changes or file modifications detected.
v1.0.3
oasis-audio v1.0.3
- Changed skill name formatting from "Oasis Audio" to "oasis-audio" for consistency.
- Added an explicit version field ("version: 1.0.3") to SKILL.md metadata.
- No changes to code or core functionality; instructions and personalized audio guidance remain the same.
- No file changes detected.
v1.0.2
- No file changes were detected in this release.
- Content, functionality, and instructions remain unchanged from the previous version.
v1.0.1
**Summary:**
Introduces advanced personalization by mining user conversation history and context to tailor audio content, including a new context_collector.py tool.
- Added `context_collector.py` for collecting and analyzing user conversation context to personalize audio.
- Skill now infers user intent, emotional state, and knowledge level from conversation and memory, not just current request.
- Supports 9 audio reference modes, but can create fully custom modes when user needs don’t fit templates.
- Audio is now tailored along four dimensions: content, tone, depth/duration, and user knowledge.
- If context mining fails or is skipped, the skill reverts seamlessly to generic audio generation.
v1.0.0
Oasis Audio 1.0.0
- Introduces AI-powered audio narration generation with 8 smart modes, including Daily Briefing, Content Digest, Bedtime Radio, and more.
- Supports both English and Chinese, producing 15–30 minute, single-narrator audio files with background music.
- Zero configuration required; just install and use without an API key.
- Automated job polling is included—status is checked and results are reported automatically.
- Offers clear guidelines for preparing and condensing user input for optimized audio generation.
Metadata
Frequently Asked Questions
What is Oasis Audio?
Oasis Audio is an AI audio narration generator that transforms your current state of mind, content you want to digest, or recent life events into a personali... It is an AI Agent Skill for Claude Code / OpenClaw, with 216 downloads so far.
How do I install Oasis Audio?
Run "/install oasis-audio" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Oasis Audio free?
Yes, Oasis Audio is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Oasis Audio support?
Oasis Audio is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Oasis Audio?
It is built and maintained by yuanyxu (@yuanyxu); the current version is v1.0.9.
More Skills