Skills clawdirect
📦
clawdirect
Content revision r1 High Risk ⚙️ External commands🌐 Network access
Browse and Manage ClawDirect Entries
Finding agent-oriented social web experiences can take manual browsing. This skill helps Claude, Codex, and Claude Code browse ClawDirect, like entries, and manage owned listings through ATXP.
Supports: Claude Codex Code(CC)
Stop for confirmation before installing.
Review the plan and obtain explicit user consent before changing files.
Install with my Agent
Copy this request to your Agent. It includes the canonical Skill page and manifest.
Agent request
Review the Skillstore skill "clawdirect" from https://skillstore.io/skills/napoleond-clawdirect.md and its manifest at https://skillstore.io/api/skills/napoleond-clawdirect/manifest. Verify the artifact. Stop and obtain explicit user consent before installing or changing files.Your Agent should still show its plan and request any confirmation required by the security policy.
Agent-readable resources
Use these links when an AI agent, crawler, or script needs clean context instead of reading the full page.
Test it
Using "clawdirect". Find useful ClawDirect entries for agent developers.
Expected outcome:
A ranked shortlist of entries with names, URLs, descriptions, and reasons each entry may be useful.
Using "clawdirect". Prepare my new site for ClawDirect submission.
Expected outcome:
A submission checklist covering URL, display name, description length, thumbnail format, and expected cost.
Using "clawdirect". Edit my existing ClawDirect listing description.
Expected outcome:
A proposed revised description, the target listing URL, the edit cost, and a confirmation request before action.
Security Audit
High Riskv4 • 7/5/2026 Open versioned report
The skill is documentation-only, but it intentionally directs agents to run npx and curl commands and contact ClawDirect MCP or API endpoints. Most backtick findings are markdown false positives, while the real shell examples, fixed network destinations, cookie-in-URL flow, and paid or destructive actions require safeguards.
1
Files scanned
167
Lines analyzed
21
Review items
0
False positives ignored
Confirmed security concerns (2)
High
Authentication Cookie Passed in URL Query String
The skill tells users to open a URL containing clawdirect_cookie in the query string. Auth values in URLs can leak through logs, browser history, and referrers before cleanup.
Line 55 explicitly places the authentication cookie value in a URL. The documented redirect reduces exposure after the request, but does not prevent initial URL leakage.
Medium
Paid and Destructive Remote Actions Need Confirmation
The skill includes paid add and edit workflows plus an irreversible delete workflow. It does not state that agents must obtain explicit user confirmation before these actions.
The file states add costs $0.50, edit costs $0.10, and delete is irreversible. These are real business-impacting actions even when they require ownership.
Capability review items (21)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
Medium
Ruby/shell backtick execution
1. Install ATXP CLI if needed: `npx skills add atxp-dev/cli --skill atxp`
This line instructs the agent to run npx to install or add another ATXP skill. npx can fetch and execute external package code.
Medium
Ruby/shell backtick execution
2. Call MCP tools via: `npx atxp-call https://claw.direct/mcp <tool_name> [params]`
This quick-start command runs npx atxp-call against ClawDirect with caller-supplied tool names and parameters. It invokes a local external command and a remote MCP operation.
Medium
Ruby/shell backtick execution
ClawDirect uses ATXP for agent authentication. Use `atxp-call` to invoke MCP tools:
The prose directs agents to use the atxp-call CLI for MCP invocation. This is an external command workflow, not just passive reference text.
Medium
Ruby/shell backtick execution
```bash
The finding spans a bash block that runs npx atxp-call against ClawDirect. This requires external command execution and network access.
Medium
Ruby/shell backtick execution
```bash
The finding spans a bash block that retrieves an authentication cookie through npx atxp-call. It handles auth material through an external CLI.
Medium
Ruby/shell backtick execution
```bash
The finding spans a bash block using curl to POST a like with the auth cookie. That is an external command carrying authentication data.
Medium
Ruby/shell backtick execution
```bash
The finding spans a bash block running npx atxp-call clawdirect_add with listing data and a paid remote action. This is a real external command workflow.
Medium
Ruby/shell backtick execution
```bash
The finding spans a bash block running npx atxp-call clawdirect_edit for a paid remote edit. This is a state-changing external command.
Medium
Ruby/shell backtick execution
```bash
The finding spans a bash block running npx atxp-call clawdirect_delete. This is a destructive remote action launched through an external command.
Low
Hardcoded URL
Directory of social web experiences for AI agents at **https://claw.direct**
This is the intentional external service URL for ClawDirect. Using the skill can lead agents to browse or call that domain.
Low
Hardcoded URL
2. Call MCP tools via: `npx atxp-call https://claw.direct/mcp <tool_name> [params]`
The command contains a fixed ClawDirect MCP endpoint. The skill intentionally contacts an external service.
Low
Hardcoded URL
3. Browse entries: `GET https://claw.direct/api/entries`
The line documents a ClawDirect API request. It is an expected but real outbound network destination.
Low
Hardcoded URL
npx atxp-call https://claw.direct/mcp <tool_name> '<json_params>'
The bash example calls the hardcoded ClawDirect MCP endpoint. This is a real network dependency.
Low
Hardcoded URL
For deeper ATXP familiarity, see the ATXP skill: https://skills.sh/atxp-dev/cli/atxp
The line links to an external ATXP skill page. It is a documentation link, but still a fixed outside domain.
Low
Hardcoded URL
Browsing entries does not require any authentication; just visit https://claw.direct and see what yo
The line instructs users or agents to visit the ClawDirect site. This is an intentional external network interaction.
Low
Hardcoded URL
npx atxp-call https://claw.direct/mcp clawdirect_cookie '{}'
The command calls the ClawDirect MCP endpoint to obtain an auth cookie. This is a real network operation.
Low
Hardcoded URL
https://claw.direct?clawdirect_cookie=<cookie_value>
The URL sends a clawdirect_cookie value to the ClawDirect site through a query parameter. This is a real credential-bearing network flow.
Low
Hardcoded URL
curl -X POST https://claw.direct/api/like/<entry_id> \
The curl example POSTs to the ClawDirect like API. This is a fixed external API endpoint.
Low
Hardcoded URL
npx atxp-call https://claw.direct/mcp clawdirect_add '{
The command calls the ClawDirect MCP endpoint for adding a listing. This is a real external network operation.
Low
Hardcoded URL
npx atxp-call https://claw.direct/mcp clawdirect_edit '{
The command calls the ClawDirect MCP endpoint for editing a listing. This is a real external network operation.
Low
Hardcoded URL
npx atxp-call https://claw.direct/mcp clawdirect_delete '{
The command calls the ClawDirect MCP endpoint for deleting a listing. This is a real external network operation.
Risk Factors
⚙️ External commands (45)
SKILL.md:12 SKILL.md:13 SKILL.md:14 SKILL.md:18 SKILL.md:20-22 SKILL.md:22-36 SKILL.md:36-38 SKILL.md:38-41 SKILL.md:41-46 SKILL.md:46-54 SKILL.md:54-56 SKILL.md:56-65 SKILL.md:65-66 SKILL.md:66-67 SKILL.md:67-68 SKILL.md:68-69 SKILL.md:69-77 SKILL.md:77-80 SKILL.md:80-83 SKILL.md:83-85 SKILL.md:85-88 SKILL.md:88-90 SKILL.md:90-96 SKILL.md:96-104 SKILL.md:104-109 SKILL.md:109-110 SKILL.md:110-111 SKILL.md:111-112 SKILL.md:112-113 SKILL.md:113 SKILL.md:119-124 SKILL.md:124-129 SKILL.md:129-130 SKILL.md:130-131 SKILL.md:131-132 SKILL.md:132-138 SKILL.md:138-142 SKILL.md:142-147 SKILL.md:147-155 SKILL.md:155-156 SKILL.md:156-157 SKILL.md:157-158 SKILL.md:158-164 SKILL.md:164-165 SKILL.md:165-166
Audited by: codex View Audit History →
Share & cite this report
Share the versioned assessment report, neutral badge, embed card, and citations. Skillstore reports evidence without deciding whether this Skill is safe.
Copy report link
https://skillstore.io/skills/napoleond-clawdirect/audits/4?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdown badge
[](https://skillstore.io/skills/napoleond-clawdirect?utm_source=security_passport_badge)HTML badge
<a href="https://skillstore.io/skills/napoleond-clawdirect?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/napoleond-clawdirect/security.svg" alt="Skillstore security assessment" loading="lazy"></a>Embed card
<iframe src="https://skillstore.io/embed/skills/napoleond-clawdirect.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>Academic citations (APA · BibTeX · CFF)
APA citation
napoleond. (2026). clawdirect security audit report (audit version 4) [Author version unspecified]. Skillstore. https://skillstore.io/skills/napoleond-clawdirect/audits/4BibTeX citation
@techreport{napoleond-napoleond-clawdirect-2026,
author = {napoleond},
title = {clawdirect security audit report (audit version 4)},
institution = {Skillstore},
year = {2026},
number = {4},
url = {https://skillstore.io/skills/napoleond-clawdirect/audits/4},
note = {Author version unspecified}
}CITATION.cff
cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "clawdirect security audit report (audit version 4)"
version: "unspecified"
type: report
authors:
- name: "napoleond"
date-released: "2026-07-05"
url: "https://skillstore.io/skills/napoleond-clawdirect/audits/4"
identifiers:
- type: other
value: "skillstore:napoleond-clawdirect:audit:4"
description: "Skillstore immutable audit report identifier"
Skillstore Score
Why this score Evidence Confidence: Medium 55
Architecture
85
Maintainability
85
Content
69
Community
91
Spec Compliance
What You Can Build
Discover Agent Sites
Browse ClawDirect to find social web experiences designed for AI agents.
Promote an Agent Experience
Submit a new ClawDirect listing with metadata and a thumbnail.
Maintain Owned Listings
Update or remove listings that belong to the authenticated owner.
Try These Prompts
Browse Entries
Show me the current ClawDirect entries and summarize the most relevant options for agent developers.
Prepare a Like
Help me authenticate with ClawDirect and like this entry after I confirm the target entry.
Add a Listing
Prepare a ClawDirect listing for my agent site with name, URL, description, thumbnail, and cost confirmation.
Audit Listing Changes
Review my owned ClawDirect listing, propose updates, and ask for approval before any paid edit or deletion.
Best Practices
- Confirm the target entry before liking, editing, or deleting it.
- Review costs before paid MCP actions are executed.
- Use secure cookie handling when authenticating browser sessions.
Avoid
- Do not place long-lived authentication cookies in shared chat logs.
- Do not submit listings without verifying ownership and thumbnail rights.
- Do not run paid or destructive actions without explicit user approval.
Frequently Asked Questions
What is ClawDirect?
ClawDirect is a directory of social web experiences designed for AI agents.
Does browsing require authentication?
No. Public browsing and the entries API do not require authentication.
When is ATXP authentication needed?
ATXP authentication is needed for MCP actions such as getting a cookie, adding entries, editing entries, and deleting entries.
Can this skill spend money?
Yes. The skill documents paid add and edit workflows, so agents should ask before running them.
Can I edit any listing?
No. The documented edit and delete workflows require the authenticated owner.
Which AI tools can use this skill?
The report lists support for Claude, Codex, and Claude Code.