Security audit
Security checks for vulnerabilities and agentic risk
This skill is a straightforward local MissionClaw project-management integration with some documentation overstatement but no hidden or destructive behavior found.
Install only if you use MissionClaw and are comfortable sending project names and descriptions to the configured MissionClaw API. Keep MISSIONCLAW_URL set to localhost or another service you trust, and treat the broader agent/scheduler features as MissionClaw app capabilities rather than commands implemented by this skill.
The skill description promises broad project management, orchestration, scheduling, and agent-management capabilities, but the documented behavior only shows simple project API usage. This mismatch is dangerous because agents or users may trust the skill with broader authority or more sensitive workflows than it actually implements, reducing transparency and increasing the chance of misuse or unsafe invocation.
The skill advertises network/API interaction with localhost and implies code-capable behavior, but it declares no explicit tool scope or permissions boundary. In an agent environment, missing least-privilege declarations can let the runtime grant broader access than intended, making unintended network or environment access harder to audit and constrain.
Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.
---
name: missionclaw
description: >
Integrates with MissionClaw for project management and AI agent orchestration. Use when the user wants to create projects, manage tasks via Kanban, assign work to agents, view organization charts, or schedule automated tasks. Calls the MissionClaw API at http://localhost:3000.
---
# MissionClaw Skill
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
Example curl:
curl -X POST http://localhost:3000/api/projects \
-H "Content-Type: application/json" \
-d '{
"projectName": "The DigiCode Landing Page",
The skill forwards raw user-supplied project details to the MissionClaw API without any disclosure, confirmation, or consent mechanism in the code path. Even though the target defaults to localhost, the endpoint is configurable via an environment variable, so user-entered content may be transmitted to another service unexpectedly, creating privacy and data-handling risk.
This manifest defines "/mc" as an activation command, but the short abbreviation is not self-describing and the file provides no constraints, exclusions, or negative examples clarifying when it should or should not be used. In a manifest, such minimal trigger naming can create ambiguity about intended invocation scope compared with the more specific "/missionclaw" alias.
Detected: suspicious.env_credential_access