tdg-personal/autonomous-agent-harness
Transform Claude Code into a fully autonomous agent system with persistent memory, scheduled operations, computer use, and task queuing. Replaces standalone agent frameworks (Hermes, AutoGPT) by leveraging Claude Code's native crons, dispatch, MCP tools, and memory. Use when the user wants continuous autonomous operation, scheduled tasks, or a self-directing agent loop.
58
73%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
View guide
Low
Low-risk findings worth noting
Low
Low-risk findings.
2 low severity findings. Worth noting, but not necessarily harmful.
W011: Third-party content exposure detected (indirect prompt injection risk).
What this means
The skill exposes the agent to untrusted, user-generated content from public third-party sources, creating a risk of indirect prompt injection. This includes browsing arbitrary URLs, reading social media posts or forum comments, and analyzing content from unknown websites.
Why it was flagged
The SKILL.md explicitly instructs the agent to fetch and act on open/public content—e.g., scheduled tasks and example workflows that "Review open PRs", "Check my GitHub notifications", "Run Exa searches", and use the "browser automation" computer-use MCP—so the agent will ingest untrusted, user-generated web content (GitHub PRs and search results) that can materially influence actions like posting reviews or creating issues.
Where we found it
github.com
domain · 7 sites
The plugin instructs the agent to fetch GitHub PRs, notifications, and issues via GitHub MCP, then act on them (review, post comments, create issues), exposing the agent to untrusted user-generated content that could contain prompt injection.
exa.ai
domain · 2 sites
The plugin instructs the agent to run Exa web searches and summarize findings, ingesting arbitrary third-party web content that could contain prompt injection payloads.
SKILL.md
43
│ │ memory github exa supabase browser-use │ │
SKILL.md
244
2. Run Exa searches for each query
browser-automated web content
content-type · 4 sites
The plugin enables computer-use MCP for browser automation that navigates, reads, and acts on arbitrary web pages, exposing the agent to untrusted web content.
SKILL.md
43
│ │ memory github exa supabase browser-use │ │
SKILL.md
132
- Browser automation (navigate, click, fill forms, screenshot)
SKILL.md
176
| Computer Use | computer-use MCP | Native browser and desktop control |
SKILL.md
224
Grant computer-use MCP the necessary permissions for browser and desktop control.
W012: Unverifiable external dependency detected (runtime URL that controls agent).
What this means
The skill fetches instructions or code from an external URL at runtime, and the fetched content directly controls the agent’s prompts or executes code. This dynamic dependency allows the external source to modify the agent’s behavior without any changes to the skill itself.
Why it was flagged
The skill's setup directs runtime execution of remote npm packages via npx (e.g., npx -y @anthropic/memory-mcp-server, npx -y @anthropic/scheduled-tasks-mcp-server, npx -y @anthropic/computer-use-mcp-server), which fetches and runs remote code as required MCP servers for the agent harness.
Where we found it
@anthropic/memory-mcp-server
dependency · 1 site
The plugin's setup instructs running npx -y @anthropic/memory-mcp-server, which fetches and executes a remote npm package at runtime without version pinning.
SKILL.md
191
"args": ["-y", "@anthropic/memory-mcp-server"]
@anthropic/scheduled-tasks-mcp-server
dependency · 1 site
The plugin's setup instructs running npx -y @anthropic/scheduled-tasks-mcp-server, which fetches and executes a remote npm package at runtime without version pinning.
SKILL.md
195
"args": ["-y", "@anthropic/scheduled-tasks-mcp-server"]
@anthropic/computer-use-mcp-server
dependency · 1 site
The plugin's setup instructs running npx -y @anthropic/computer-use-mcp-server, which fetches and executes a remote npm package at runtime without version pinning.
SKILL.md
199
"args": ["-y", "@anthropic/computer-use-mcp-server"]
- Audited
- Security analysis