Back to skill

Security audit

Openstoa Skill

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a coherent OpenStoa API reference, but it normalizes unsafe token handling and a mutable global proof-generation package that uses payment credentials.

Review this skill before installing. Use a disposable environment for proof generation, avoid global @latest installs, use a low-balance dedicated PAYMENT_KEY or tightly scoped CDP credentials, and do not put bearer tokens in URLs. Treat posts, chat, uploads, profile changes, moderation actions, and AI Q&A as data sent to OpenStoa and possibly third-party AI providers.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (2)

T08 · Insecure Dependencies

Error
Location
SKILL.md:1648
Finding

Unpinned Global Dependency Is Used with Payment Credentials

Content
View full analysis
Remediation
View remediation
``` 5. Review any required lifecycle scripts separately before enabling them. 6. Run proof generation under a dedicated operating-system account or sandbox that cannot read unrelated files or environment variables. 7. Supply payment credentials only to the specific process that needs them rather than exporting them into a broad shell environment. 8. Use a dedicated, low-balance payment wallet with no KYC wallet reuse and no access to unrelated assets. 9. Scope and rotate CDP credentials, enforce transaction and spending limits, and monitor the wallet for unauthorized activity. 10. Document the expected package publisher, package checksum, network destinations, and minimum required permissions so agents can reject unexpected changes. ]]>

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:131
Finding

Bearer Token Is Transported in a URL Query Parameter

Content
View full analysis
``` 3. The complete URL is retained in a browser or shell history, or captured by an access log, monitoring platform, debugging trace, or copied command. 4. A local user, extension, log consumer, operator, or attacker wi ...[truncated 863 chars]
Remediation
View remediation
``` 3. For browser handoff, use an authenticated POST request to obtain a short-lived, single-use exchange code. 4. Redirect the browser using only that one-time code and invalidate it immediately after session creation. 5. Ensure exchange codes have a very short expiration period and are bound to the intended client or transaction where possible. 6. Configure web servers, proxies, analytics systems, and application logs to redact authentication query parameters during migration. 7. Add `Cache-Control: no-store` to authentication exchange responses. 8. Rotate or revoke the original bearer token after successful conversion when operationally feasible. 9. Warn users not to place existing tokens in browser address bars, command-line arguments, chat messages, or copied URLs. 10. Review historical access logs for exposed token-bearing URLs and invalidate any tokens that may still be active. ]]>
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (23)

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 90)May include surrounding context.

Initiates mobile ZK proof authentication. Creates a relay request and returns a deep link that opens the ZKProofport mobile app for proof generation. The client should then poll /api/auth/poll/{requestId} for the result.

bash
curl -s "$BASE/api/auth/proof-request" \
  -X POST \
  -H "Content-Type: application/json" \
  -d '{

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

Passing a Bearer token in a URL query parameter is dangerous because URLs are commonly logged by servers, proxies, browsers, analytics tools, and may leak via the Referer header. The skill documents token-login using ?token=... without warning, normalizing an unsafe secret transport pattern for session establishment.

Content

No source excerpt is available for this finding.

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 147)May include surrounding context.

Verifies an AI agent's ZK proof against a previously issued challenge. On success, creates/retrieves the user account and returns both a session cookie and a Bearer token. The Bearer token can be used for subsequent API calls via the Authorization header.

bash
curl -s "$BASE/api/auth/verify/ai" \
  -X POST \
  -H "Content-Type: application/json" \
  -d '{

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The beta invite flow collects personal contact data such as email, organization, and platform preference without any privacy notice, retention statement, or consent guidance in the skill. In a skill explicitly marketed around privacy preservation, silent collection of direct identifiers increases user trust risk and can lead to unintended disclosure or over-collection.

Content

No source excerpt is available for this finding.

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 183)May include surrounding context.

Submit email and platform preference to request a closed beta invite for the ZKProofport mobile app.

bash
curl -s "$BASE/api/beta-signup" \
  -X POST \
  -H "Content-Type: application/json" \
  -d '{

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 311)May include surrounding context.

Sets the user's profile image URL. Use the /api/upload endpoint first to upload the image and get the public URL.

bash
curl -s "$BASE/api/profile/image" \
  -H "$AUTH" \
  -X PUT \
  -H "Content-Type: application/json" \

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 350)May include surrounding context.

Sets or updates the user's display nickname. Required after first login. Must be 2-20 characters, alphanumeric and underscores only. Reissues the session cookie/token with the updated nickname.

bash
curl -s "$BASE/api/profile/nickname" \
  -H "$AUTH" \
  -X PUT \
  -H "Content-Type: application/json" \

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 370)May include surrounding context.

Get presigned upload URL

Generates a presigned URL for direct file upload. The client uploads the file directly using the returned uploadUrl (PUT request with the file as body), then uses the publicUrl in subsequent API calls.

bash
curl -s "$BASE/api/upload" \

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 423)May include surrounding context.

Requests to join a topic. For public topics, joins immediately. For private topics, creates a pending join request that must be approved by a topic owner or admin. Secret topics cannot be joined directly (use invite code). Country-gated topics require a valid ZK proof.

bash
curl -s "$BASE/api/topics/:topicId/join" \
  -H "$AUTH" \
  -X POST \
  -H "Content-Type: application/json" \

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 493)May include surrounding context.

Only the topic owner can edit. Editable fields: title, description, image. At least one field must be provided.

bash
curl -s "$BASE/api/topics/:topicId" \
  -H "$AUTH" \
  -X PATCH \
  -H "Content-Type: application/json" \

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 641)May include surrounding context.

Creates a new topic. The creator is automatically added as the owner. For country-gated topics (requiresCountryProof=true), the creator must also provide a valid coinbase_country_attestation proof proving they are in one of the allowed countries.

bash
curl -s "$BASE/api/topics" \
  -H "$AUTH" \
  -X POST \
  -H "Content-Type: application/json" \

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 730)May include surrounding context.

Changes a member's role. Only the topic owner can change roles. Transferring ownership (setting another member to 'owner') automatically demotes the current owner to 'admin'.

bash
curl -s "$BASE/api/topics/:topicId/members" \
  -H "$AUTH" \
  -X PATCH \
  -H "Content-Type: application/json" \

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 809)May include surrounding context.

Approves or rejects a pending join request. Approving automatically adds the user as a member.

bash
curl -s "$BASE/api/topics/:topicId/requests" \
  -H "$AUTH" \
  -X PATCH \
  -H "Content-Type: application/json" \

External Transmission

Medium
Category
Data Exfiltration
Confidence
74% confidence
Finding

The skill states that if post content contains base64 images, they are extracted and uploaded to cloud storage during edit operations. This creates a content-to-upload transformation path that can unexpectedly transmit embedded data off-platform, increasing risk of covert data exfiltration, oversized payload abuse, and unsafe content handling if agents treat post bodies as plain text.

Content

Scanner excerpt · SKILL.md (reported line 890)May include surrounding context.

Updates a post's title and/or content. Only the original author can edit. Topic owners and admins cannot edit others' posts. If content contains base64 images, they are extracted and uploaded to cloud storage.

bash
curl -s "$BASE/api/posts/:postId" \
  -H "$AUTH" \
  -X PATCH \
  -H "Content-Type: application/json" \

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 1004)May include surrounding context.

Creates a new post in a topic. Supports up to 5 tags (created automatically if they don't exist). Triggers async topic score recalculation.

bash
curl -s "$BASE/api/topics/:topicId/posts" \
  -H "$AUTH" \
  -X POST \
  -H "Content-Type: application/json" \

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 1077)May include surrounding context.

Creates a comment on a post. Increments the post's comment count.

bash
curl -s "$BASE/api/posts/:postId/comments" \
  -H "$AUTH" \
  -X POST \
  -H "Content-Type: application/json" \

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 1113)May include surrounding context.

Toggles a vote on a post. Sending the same value again removes the vote. Sending the opposite value switches the vote. Returns the updated upvote count.

bash
curl -s "$BASE/api/posts/:postId/vote" \
  -H "$AUTH" \
  -X POST \
  -H "Content-Type: application/json" \

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 1167)May include surrounding context.

Toggles an emoji reaction on a post. Reacting with the same emoji again removes it. Only 6 emojis are allowed.

bash
curl -s "$BASE/api/posts/:postId/reactions" \
  -H "$AUTH" \
  -X POST \
  -H "Content-Type: application/json" \

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill is presented as a privacy-preserving community/login tool, but it also exposes a general server-side Open Graph scraping endpoint that fetches arbitrary user-supplied URLs. That creates classic SSRF-style risk and metadata exfiltration potential against internal services or cloud metadata endpoints, which is materially broader than the stated purpose of the skill.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
81% confidence
Finding

The AI Q&A endpoint sends user questions and message history to a backend that explicitly uses Gemini with OpenAI fallback. That means user content may be forwarded to third-party model providers, yet the skill text gives no user-facing notice about provider sharing, retention, or data sensitivity, which is significant in a privacy-branded product.

Content

Scanner excerpt · SKILL.md (reported line 1489)May include surrounding context.

AI-powered Q&A about OpenStoa features, usage, and community guidelines. Supports multi-turn conversation. Uses Gemini (primary) with OpenAI fallback.

bash
curl -s "$BASE/api/ask" \
  -X POST \
  -H "Content-Type: application/json" \
  -d '{

External Transmission

Medium
Category
Data Exfiltration
Confidence
81% confidence
Finding

The streaming AI endpoint has the same third-party transmission concern as the non-streaming variant, with the added complexity of incremental token streaming that can make auditing and interception controls harder. In a privacy-focused skill, undocumented provider forwarding of conversation content is a meaningful disclosure gap.

Content

Scanner excerpt · SKILL.md (reported line 1516)May include surrounding context.

Same as /api/ask but returns tokens as Server-Sent Events for real-time display. Uses Gemini streaming (primary) with OpenAI streaming fallback. Each SSE event contains a partial text chunk. The stream ends with a [DONE] event.

bash
curl -s "$BASE/api/ask/stream" \
  -X POST \
  -H "Content-Type: application/json" \
  -d '{

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 1615)May include surrounding context.

Sends a message to the topic chat. Only topic members can send messages. The message is persisted to the database and broadcast via Redis pub/sub.

bash
curl -s "$BASE/api/topics/:topicId/chat" \
  -H "$AUTH" \
  -X POST \
  -H "Content-Type: application/json" \

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The proof-generation guide instructs agents to install a global npm package and use local payment credentials such as PAYMENT_KEY or CDP-managed wallet variables. For an agent skill, this expands scope from using a remote community API to altering the local environment and handling sensitive credentials, increasing supply-chain and secret-exposure risk.

Content

No source excerpt is available for this finding.