Back to skill

Security audit

voidly

Security checks for vulnerabilities and agentic risk

Overview

The scanners classify this skill as benign because its reported external connections align with its stated purpose, though overall confidence is medium and coverage is incomplete.

Before installing, consider that using the described features involves contacting api.voidly.ai and that watch or identity features involve an agent key. The verdict is benign, not risk-free; the scan was partial because references to a local key-header file could not be resolved, so those details were not fully assessed.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (15)

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 3)May include surrounding context.

md
---
name: voidly
description: Read Voidly's internet-censorship index answer for one country, search recent incident records with evidence links, follow new records for a country, domain or incident type through a pull feed, and start a self-owned Voidly relay identity whose keys stay on the local machine. Uses plain curl GETs or the hosted MCP server at api.voidly.ai/mcp. Use when someone asks what Voidly shows for a country, which censorship or disruption incidents it has recorded, or wants to follow new incident records.
homepage: https://voidly.ai/agents
metadata:
  openclaw:

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 32)May include surrounding context.

1. Check one country

bash
curl -s https://api.voidly.ai/v1/answers/country/IR
curl -s "https://api.voidly.ai/v1/answers/country/IR?format=json"

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 33)May include surrounding context.

bash
curl -s https://api.voidly.ai/v1/answers/country/IR
curl -s "https://api.voidly.ai/v1/answers/country/IR?format=json"

Use ISO 3166-1 alpha-2 codes (IR, RU, CN, PK). The plain-text answer is ready to quote. It gives a 7-day OONI anomaly signal: a score 0-100, a level, a rank, the measurement count and a data-as-of time, then the newest citable incidents and a list of what the answer cannot tell you.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 47)May include surrounding context.

2. Search incidents

bash
curl -s "https://api.voidly.ai/data/incidents?country=IR&limit=5"
curl -s "https://api.voidly.ai/data/incidents?country=IR&limit=5&citable=true"
curl -s https://api.voidly.ai/data/incidents/IR-2026-2534/evidence

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 48)May include surrounding context.

bash
curl -s "https://api.voidly.ai/data/incidents?country=IR&limit=5"
curl -s "https://api.voidly.ai/data/incidents?country=IR&limit=5&citable=true"
curl -s https://api.voidly.ai/data/incidents/IR-2026-2534/evidence

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 49)May include surrounding context.

bash
curl -s "https://api.voidly.ai/data/incidents?country=IR&limit=5"
curl -s "https://api.voidly.ai/data/incidents?country=IR&limit=5&citable=true"
curl -s https://api.voidly.ai/data/incidents/IR-2026-2534/evidence

Each record has id (readable, like IR-2026-2534), incidentType (censorship, mixed or disruption), status, sources, startTime, createdAt and reportUrl.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 77)May include surrounding context.

md
The watch tools read the agent key only from the `X-Agent-Key` HTTP header of the MCP request, never from a tool argument. To use them, add `"headers": {"X-Agent-Key": "${VOIDLY_AGENT_KEY}"}` to the server entry so OpenClaw fills it from the environment.

Alternative: an A2A v0.3.0 JSON-RPC endpoint at `https://api.voidly.ai/v1/a2a`, card at `https://api.voidly.ai/v1/a2a/agent-card.json`.

## 4. Start an identity

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 116)May include surrounding context.

md
- Write "Voidly's index shows ..." with the date. Do not present a score as a verdict, a block list or a forecast.
- Do not state that a site is reachable or blocked now.
- Data license: Voidly-original data CC BY 4.0; OONI-derived raw data CC BY-NC-SA 4.0. Credit Voidly and OONI.
- Abuse policy: https://api.voidly.ai/v1/abuse/policy
- On HTTP 429, wait for Retry-After. On 503, nothing was created; try again later.
- More for agents: https://voidly.ai/agents

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 62)May include surrounding context.

3. Hosted MCP (optional)

Server: https://api.voidly.ai/mcp (Streamable HTTP). Public reads need no key. Server entry for OpenClaw:

json
{"url": "https://api.voidly.ai/mcp", "transport": "streamable-http"}

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 65)May include surrounding context.

Server: https://api.voidly.ai/mcp (Streamable HTTP). Public reads need no key. Server entry for OpenClaw:

json
{"url": "https://api.voidly.ai/mcp", "transport": "streamable-http"}

Save it with openclaw mcp set voidly '<the JSON above>'. Useful tools:

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 93)May include surrounding context.

5. Follow incidents with a watch

bash
curl -s -X POST https://api.voidly.ai/v1/agent/watches -H @"$HOME/.voidly/agent-key.header" -H "Content-Type: application/json" -d '{"country_code":"IR"}'
curl -s "https://api.voidly.ai/v1/agent/watches/feed?cursor=0" -H @"$HOME/.voidly/agent-key.header"
curl -s https://api.voidly.ai/v1/agent/watches -H @"$HOME/.voidly/agent-key.header"

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 77)May include surrounding context.

md
The watch tools read the agent key only from the `X-Agent-Key` HTTP header of the MCP request, never from a tool argument. To use them, add `"headers": {"X-Agent-Key": "${VOIDLY_AGENT_KEY}"}` to the server entry so OpenClaw fills it from the environment.

Alternative: an A2A v0.3.0 JSON-RPC endpoint at `https://api.voidly.ai/v1/a2a`, card at `https://api.voidly.ai/v1/a2a/agent-card.json`.

## 4. Start an identity

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 94)May include surrounding context.

bash
curl -s -X POST https://api.voidly.ai/v1/agent/watches -H @"$HOME/.voidly/agent-key.header" -H "Content-Type: application/json" -d '{"country_code":"IR"}'
curl -s "https://api.voidly.ai/v1/agent/watches/feed?cursor=0" -H @"$HOME/.voidly/agent-key.header"
curl -s https://api.voidly.ai/v1/agent/watches -H @"$HOME/.voidly/agent-key.header"

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 95)May include surrounding context.

bash
curl -s -X POST https://api.voidly.ai/v1/agent/watches -H @"$HOME/.voidly/agent-key.header" -H "Content-Type: application/json" -d '{"country_code":"IR"}'
curl -s "https://api.voidly.ai/v1/agent/watches/feed?cursor=0" -H @"$HOME/.voidly/agent-key.header"
curl -s https://api.voidly.ai/v1/agent/watches -H @"$HOME/.voidly/agent-key.header"
  • Scope: exactly one of country_code, domain or incident_type (censorship, mixed, disruption); optional min_severity.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 83)May include surrounding context.

md
Only when the person asks for a Voidly identity, a relay inbox or a watch.

1. Read the protocol: `curl -s https://api.voidly.ai/v1/agents/start`. It returns the request format, limits, what the server can read and a dependency-free Node 20 example.
2. Read and review the whole example (`example_node`) before running anything. Confirm it contacts only api.voidly.ai and prints no key. It holds keys in memory only, so add a step that saves the API key and both private keys to a local file readable only by the person, and one that writes the single line `X-Agent-Key: <the API key>` to `~/.voidly/agent-key.header` (readable only by the person). Set its `watch` line to the scope wanted (or remove it).
3. The example generates an Ed25519 signing key, an X25519 encryption key and a random API key locally. The server receives only the two public keys and the SHA-256 of the API key, in one signed POST to `/v1/agents/start` (Ed25519 signature plus a 20-bit proof of work, about a second of local