Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed Outgoing integration for finding and booking activities, with expected third-party API calls and clear purchase-confirmation guidance.
Install this only if you are comfortable sending outing searches, location hints, and a stable external user ID to Outgoing. Use dry-run payment tokens while testing, and only provide a real payment token after confirming the exact tickets, quantity, and total cost.
The manifest description says to use this skill for any "what should we do / where should we eat / what's on near me" request. These are common conversational phrases and the file does not provide clear exclusion conditions or boundaries for when this general skill should not activate beyond a loose suggestion to route to narrower companion skills.
The trigger list includes the standalone term "accessible," which is common everyday language and not specific enough to this skill's domain. In a markdown skill description, such a broad trigger can cause unintended invocation for unrelated requests involving general accessibility or usability.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
env:
- OUTGOING_API_KEY
anyBins:
- curl
envVars:
- name: OUTGOING_API_KEY
required: true
The listed triggers include generic single words like "nature," "outdoors," "park," and "garden," plus a broad intent statement that the user "wants the outdoors and fresh air." In a markdown skill description, these vague triggers can match many ordinary conversations and the file does not provide exclusion conditions or narrower activation constraints.
The skill instructs the agent to transmit user prompts and a persistent external user identifier to a third-party API, but it does not clearly warn that user-related data will leave the local system. This creates a meaningful privacy and consent risk, especially because free-form prompts may contain sensitive location, preference, or behavioral information.
This skill is explicitly designed to call an external API and transmit natural-language prompts plus X-External-User-Id, so external transmission is real and not hypothetical. In context this is expected functionality, but it still carries privacy and data-handling risk because user interests, whereabouts, and identifiers may be disclosed to a third party.
curl -sS -G https://api.outgoing.world/partner/v1/search \
-H "Authorization: Bearer $OUTGOING_API_KEY" \
-H "X-External-User-Id: user-123" \
--data-urlencode "prompt=scenic outdoor walks and gardens to relax this weekend" \
The trigger guidance is broad enough to match common conversational phrases like 'with friends' or 'happy hour', which can cause the skill to activate when the user did not clearly intend to invoke this specific external-search capability. That increases the chance of unintended routing and downstream data disclosure to the external Outgoing service, especially because the skill is designed to send natural-language prompts about user plans and location context.
The skill instructs the agent to provision users and send an external_user_id in the X-External-User-Id header on every request, but it does not mention any user notice, consent flow, or data-handling constraints. Sending stable user identifiers to a third-party API without transparency or minimization creates privacy and tracking risk, particularly when combined with behavioral context such as nightlife interests and city/neighborhood searches.
The skill lists triggers such as "pet-friendly" and "somewhere with my pet," which are broad natural-language phrases that may overlap with casual user requests rather than a clearly bounded invocation condition. The description does not provide exclusion conditions or negative examples to clarify when the skill should not activate.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -sS -G https://api.outgoing.world/partner/v1/search \
-H "Authorization: Bearer $OUTGOING_API_KEY" \
-H "X-External-User-Id: user-123" \
--data-urlencode "prompt=dog-friendly cafés and patios where I can bring my dog this weekend" \
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -sS -G https://api.outgoing.world/partner/v1/search \
-H "Authorization: Bearer $OUTGOING_API_KEY" \
-H "X-External-User-Id: user-123" \
--data-urlencode "prompt=dog-friendly cafés and patios where I can bring my dog this weekend" \
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -sS -G https://api.outgoing.world/partner/v1/search \
-H "Authorization: Bearer $OUTGOING_API_KEY" \
-H "X-External-User-Id: user-123" \
--data-urlencode "prompt=dog-friendly cafés and patios where I can bring my dog this weekend" \
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -sS -G https://api.outgoing.world/partner/v1/search \
-H "Authorization: Bearer $OUTGOING_API_KEY" \
-H "X-External-User-Id: user-123" \
--data-urlencode "prompt=dog-friendly cafés and patios where I can bring my dog this weekend" \
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -sS -G https://api.outgoing.world/partner/v1/search \
-H "Authorization: Bearer $OUTGOING_API_KEY" \
-H "X-External-User-Id: user-123" \
--data-urlencode "prompt=dog-friendly cafés and patios where I can bring my dog this weekend" \
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -sS -G https://api.outgoing.world/partner/v1/search \
-H "Authorization: Bearer $OUTGOING_API_KEY" \
-H "X-External-User-Id: user-123" \
--data-urlencode "prompt=dog-friendly cafés and patios where I can bring my dog this weekend" \
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -sS -G https://api.outgoing.world/partner/v1/search \
-H "Authorization: Bearer $OUTGOING_API_KEY" \
-H "X-External-User-Id: user-123" \
--data-urlencode "prompt=dog-friendly cafés and patios where I can bring my dog this weekend" \
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -sS -G https://api.outgoing.world/partner/v1/search \
-H "Authorization: Bearer $OUTGOING_API_KEY" \
-H "X-External-User-Id: user-123" \
--data-urlencode "prompt=dog-friendly cafés and patios where I can bring my dog this weekend" \
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -sS -G https://api.outgoing.world/partner/v1/search \
-H "Authorization: Bearer $OUTGOING_API_KEY" \
-H "X-External-User-Id: user-123" \
--data-urlencode "prompt=dog-friendly cafés and patios where I can bring my dog this weekend" \
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -sS -G https://api.outgoing.world/partner/v1/search \
-H "Authorization: Bearer $OUTGOING_API_KEY" \
-H "X-External-User-Id: user-123" \
--data-urlencode "prompt=dog-friendly cafés and patios where I can bring my dog this weekend" \
The listed triggers include generic terms like "active" and "sporty" plus the phrase "something to get moving," which are broad enough to match ordinary conversation outside a narrow skill-invocation context. The file does not provide exclusion conditions or negative examples to clarify when the skill should not activate.
The skill sends user prompts and location context such as city, and optionally lat/lng, to an external API without any user-facing disclosure in the skill description. This creates a privacy and consent risk because users may unknowingly share sensitive behavioral or location information with a third party.
This skill explicitly performs external network transmission to api.outgoing.world and includes user query content plus a user identifier header. In the context of a recommendation skill, external calls are expected, but they still introduce third-party data-sharing, tracking, and privacy risks if users are not informed and if identifiers are not carefully managed.
curl -sS -G https://api.outgoing.world/partner/v1/search \
-H "Authorization: Bearer $OUTGOING_API_KEY" \
-H "X-External-User-Id: user-123" \
--data-urlencode "prompt=active things to do this weekend — climbing, cycling, or a drop-in class" \
The 'When to use' section lists broad phrases like 'visiting', 'what should I see', and 'itinerary' without clear constraints or negative examples. In a markdown skill description, these vague triggers could match ordinary conversation and cause unintended activation.