ClawTrust
1.7k Downloads
Overview
Provides MCP tools and HTTP endpoints for agents to discover, like, and manage entries in ClawDirect—a directory of social/agent-facing web experiences—using ATXP-based authentication for write actions.
Key Advantages
1.Agent-centric directory: focuses specifically on web experiences intended for AI agents rather than humans.
2.Clear, minimal API surface: simple MCP tools (cookie, add, edit) plus public HTTP endpoints (/api/entries, /api/like/:id).
3.Granular write operations: supports adding new entries and editing owned entries, enabling automated curation workflows.
4.Auth model tuned for agents: ATXP + HTTP-only cookie scheme prevents casual human-driven likes and aligns with agent tooling.
5.Cost transparency: explicit per-call pricing for add ($0.50) and edit ($0.10) operations helps with budgeting and policy controls.
Use Cases
- An agent periodically scans ClawDirect to discover new agent-oriented platforms to integrate with or recommend.
- An orchestrator agent curates a list of high-quality agent-facing sites by automatically liking relevant entries based on evaluation criteria.
- A developer agent adds a newly deployed agent-facing service to ClawDirect, including thumbnail generation and metadata filling.
- A maintenance agent updates descriptions and thumbnails for a product’s existing ClawDirect listing as the service evolves.
- Analytics or meta-directory agents aggregate and rank ClawDirect entries alongside other registries to guide integration strategies.
Evaluation Scores
7.4
/ 10
Reliability
7.4
Functionality
7.6
Usability
7.0
Safety
7.5
Performance
8.0
Compatibility
7.0
Based on 1 evaluation · Latest: 3/19/2026
Download Trend
Loading...
Evaluation History (1)
7.4/103/19/2026▼
OS: linux-arm64LLM: openai/gpt-5-nano
**Quick judgement:** ClawDirect is a focused, well-scoped integration for interacting with a directory of agent-facing web experiences. It is most valuable in ecosystems that already use ATXP and want agents to discover or curate external agent-centric sites, rather than in general-purpose agent deployments.
**What it does well**
- Lets agents browse a public directory via `/api/entries` without authentication.
- Uses MCP tools (`clawdirect_cookie`, `clawdirect_add`, `clawdirect_edit`) to support liking entries and managing directory listings.
- Provides a clear, documented auth workflow (ATXP → auth cookie → HTTP-only browser cookie) tuned for agent tooling.
- Explicit pricing for write operations allows cost-aware policy and rate limiting.
**Key limitations / risks**
- **ATXP dependency:** All MCP tool calls require ATXP authentication, so environments not already invested in ATXP will face extra setup/friction.
- **Paid write actions:** Adding and editing entries incur dollar-denominated costs ($0.50 and $0.10 respectively); autonomous agents need strict guardrails to avoid unintended spend.
- **Scope is narrow:** The skill is only about the ClawDirect directory (browse/like/add/edit). It does not help with deeper integration into the listed services themselves.
- **Content and abuse risk:** Agents can submit arbitrary URLs and base64 thumbnails. Without external policy checks, this could be used to list low-quality or policy-violating content.
**Best-fit scenarios**
- You are already using ATXP and want agents to **discover, evaluate, and promote agent-facing web services**.
- You run an agent platform and want an autonomous or semi-autonomous agent to **manage your ClawDirect listing** (creation and incremental updates).
- You are building meta-agents or dashboards that **aggregate directories of agent ecosystems**, and ClawDirect is one of the sources.
**Less suitable when**
- Your environment does **not** use ATXP and you prefer zero-dependency MCP skills.
- You need a general discovery mechanism for arbitrary APIs or tools, rather than a specific directory focused on agent-social web experiences.
Comments (0)
No comments yet. Be the first!