What does this skill do, and when should you use it?
This skill is for OpenCLI users. When an opencli command fails because a website changed its DOM, API, or response schema, the skill automatically diagnoses the issue, patches the adapter, and retries. It handles common error codes like SELECTOR, EMPTY_RESULT, API_ERROR, and TIMEOUT, while enforcing clear safety boundaries—it stops on authentication required or CAPTCHA issues. The process involves collecting a failure trace, analyzing the root cause, exploring the live site, applying minimal patches, and verifying with retry, up to 3 rounds. After a successful fix, it prepares an upstream GitHub issue and, with user approval, files it via gh.
Runs the failing opencli command with trace retention; reads trace summary and adapter source to diagnose errors; uses opencli browser to inspect live DOM and network activity; applies minimal patches to adapter source (e.g., update selectors, endpoints, wait conditions); re-runs command to verify; if verified, drafts an issue and, upon approval, creates it via gh issue create.
- When opencli zhihu hot fails with a SELECTOR error because the .HotList-item class changed, it finds the new class name and fixes the adapter.
- When opencli xiaohongshu search returns EMPTY_RESULT, it first checks if it's a temporary anti-scrape block before deciding to repair.
- When an API endpoint moves, causing opencli bilibili commands to fail, it discovers the new endpoint and updates the adapter.
- When a site redesign changes loading behavior, causing opencli hackernews timeouts, it adjusts wait conditions to match.
- After a successful local fix, it generates a GitHub issue template and submits it to the jackwener/OpenCLI repo with user consent.
How do you install this skill?
- Some features depend on overseas services (e.g., GitHub) that may be unreachable in mainland China, affecting usability.
- Publisher identity unverified; maintenance responsibility unclear.
- Security restrictions exist but no guidance on handling sensitive data (e.g., login credentials) from browser sessions.
- Documented repair operations may not always resolve real issues; static review can't verify efficacy.
- Shell / CLI
- Local filesystem
opencligh
Install via the repository's skill collection: run npx skills add jackwener/opencli --skill opencli-autofix. This skill assumes you have OpenCLI installed and optionally the GitHub CLI (gh) for filing issues.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/jackwener/OpenCLI.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/opencli-autofix" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- opencli zhihu hot is returning empty — fix it
Trigger with a prompt like: "opencli zhihu hot is returning empty — fix it". The skill will guide the agent through prerequisites (opencli doctor), trace collection, analysis, patching, verification, and issue filing.
What are this skill's strengths and limitations?
- Automates repair of common adapter failures, saving manual effort
- Clear safety boundaries prevent modification of core code or auth-related files
- Provides detailed error classification and repair strategies
- Automatically generates upstream issues to benefit the community
- Uses failure traces for evidence-based debugging
- Specific to OpenCLI ecosystem; not applicable to other CLI tools
- Issue filing depends on GitHub CLI and network; works only if gh is authenticated
- Cannot handle major site redesigns requiring full adapter rewrite
- No official test suite; effectiveness depends on real-world usage
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| OpenCLI AutoFix — Automatic Adapter Repair this page | 58 · Recommended | ★ 30k | 17d ago | Apache-2.0 |
| playwright-cli Browser Automation Skill | 56 · Use with care | ★ 11k | 13d ago | MIT |
| Playwright Browser Automation ✓ Microsoft · Official | 46 · Use with care | ★ 97k | 3d ago | Apache-2.0 |
| OpenCLI Smart Search Router | 41 · Not recommended | ★ 30k | 17d ago | Apache-2.0 |
| OpenCLI Adapter Authoring Guide | 56 · Use with care | ★ 30k | 17d ago | Apache-2.0 |
How did FollowSkills review this skill?
The skill shows clear hard-stop conditions (AUTH_REQUIRED, BROWSER_CONNECT, CAPTCHA/rate-limit) and restricts modification to adapterSourcePath, not touching src/, extension/, tests/, package.json etc. Allowed upstream issue filing requires explicit user confirmation. These show good safety awareness. However, there's no explicit guidance on sensitive data handling for browser sessions or downloaded content, and dependencies on external services (e.g., GitHub for issues) are not fully internal. Hence deductions, giving 17.
Documentation is self-consistent, steps clear, provides error code classification, repair strategies, examples. However, this is static review, no execution verification. Repository has CI and tests, but specific tests for this skill's key paths not clear. Hence down to 8, which is below 10 cap but above 5 lower bound.
Clear trigger conditions (specific error codes) and detailed non-use scenarios (non-adapter issues, reproducible noise). Scenario clear, intended user (AI agent) clear. However, environment fit: though targeting Chinese platforms, relies on GitHub etc. possibly unreachable from mainland China, not explicitly noted. So 12.
Well-structured doc with safety boundaries, prerequisites, when to use, step-by-step, common fixes, stop conditions, example session. Lack versioning or changelog, maintenance responsibility unclear (Publisher unverified), known limitations section missing. So 12.
Core task well-defined, but static review cannot verify actual efficacy. No real successful fixes or reproducible evidence provided. Hence 6 (<7 anchor).
No verifiable third-party execution evidence, e.g., real CI runs for this skill. Documentation is self-claimed, no independent verification. Hence 3.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →