← Back to Skills Marketplace
87
Downloads
0
Stars
1
Active Installs
1
Versions
Install in OpenClaw
/install opencli-autofix
Description
Automatically fix broken OpenCLI adapters when commands fail. Load this skill when an opencli command fails — it guides you through diagnosing the failure vi...
Usage Guidance
This skill is coherent with its purpose but exercises powerful capabilities (editing files and posting to GitHub). Before installing or enabling it: (1) require an explicit, step-by-step confirmation prompt before any local file is modified or any GitHub action is taken; (2) verify the RepairContext.adapter.sourcePath points to the intended adapter file (not a sensitive system path); (3) inspect the diagnostic JSON yourself before allowing edits; (4) do not allow automatic publication of full adapter source to public issues — scrub secrets and review diffs first; (5) ensure your gh CLI token has limited scope, and be prepared to revoke it if unintended publishing occurs; (6) consider running the skill in a sandbox or on a fork/clone rather than on production files. These mitigations would reduce the risk of accidental modification or exfiltration.
Capability Analysis
Type: OpenClaw Skill
Name: opencli-autofix
Version: 1.0.0
The 'opencli-autofix' skill enables an AI agent to automatically diagnose and patch failing OpenCLI adapters by editing source code and executing shell commands. While the instructions in SKILL.md include safety boundaries and limit modifications to specific adapter paths, the capability to modify and execute code based on external web content (DOM snapshots) is a high-risk behavior that could be targeted by indirect prompt injection. No evidence of intentional malice or data exfiltration was found, but the broad permissions (Bash, Edit, Write) and the nature of self-modifying code warrant a suspicious classification.
Capability Assessment
Purpose & Capability
Name and description align with the instructions: the skill reads an OPENCLI_DIAGNOSTIC payload, inspects adapter source, edits the adapter at the provided sourcePath, retries commands, and files GitHub issues. It does not request unrelated credentials, packages, or install steps; the allowed tools (opencli, gh, Read/Edit/Write) are relevant to the stated task.
Instruction Scope
The runtime instructions explicitly tell the agent to extract RepairContext.adapter.sourcePath from diagnostic output and edit that file. This is coherent for a repair tool, but it gives the agent the ability to modify local files and to construct GitHub issues containing adapter code. The SKILL.md does not mandate a clear, required user confirmation step before making edits or before publishing code upstream. If an attacker can influence the diagnostic JSON or the reported sourcePath, the agent could be directed to overwrite arbitrary files or leak code.
Install Mechanism
Instruction-only skill with no install spec and no code files — nothing gets downloaded or written to disk by an installer. Lowest install risk.
Credentials
The skill requests no environment variables, which matches the registry metadata. However, it instructs use of the gh CLI (Bash(gh:*)) to file upstream issues — that implies the agent will rely on whatever GH authentication is available in the environment. Creating public GitHub issues or PRs containing adapter source can leak proprietary or sensitive logic; the need for GH access is proportionate to filing issues but still security-sensitive and should be gated by explicit user consent.
Persistence & Privilege
always:false (good) but model invocation is allowed (default). Because the skill directs file edits and network actions, autonomous invocation increases risk: an agent could perform repairs and publish data without clear manual confirmation. The combination of file-write permission and downstream publishing (gh) raises the privilege/impact level.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install opencli-autofix - After installation, invoke the skill by name or use
/opencli-autofix - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Changelog for opencli-autofix v1.0.0
- Initial release of automatic OpenCLI adapter repair skill.
- Guides user through diagnosis with OPENCLI_DIAGNOSTIC, patching, and validated retries.
- Enforces hard safety boundaries (prevents unsafe file edits, stops for login or connectivity errors).
- Limits to patching only the reported adapter source file, with a maximum of three repair attempts per failure.
- Provides step-by-step troubleshooting, including DOM/API diagnostics, network inspection, and patch guidance.
- Supports any AI agent with Bash, Read, Edit, and Write tool permissions.
Metadata
Frequently Asked Questions
What is opencli-autofix?
Automatically fix broken OpenCLI adapters when commands fail. Load this skill when an opencli command fails — it guides you through diagnosing the failure vi... It is an AI Agent Skill for Claude Code / OpenClaw, with 87 downloads so far.
How do I install opencli-autofix?
Run "/install opencli-autofix" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is opencli-autofix free?
Yes, opencli-autofix is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does opencli-autofix support?
opencli-autofix is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created opencli-autofix?
It is built and maintained by chang (@liberalchang); the current version is v1.0.0.
More Skills