Back to skill

Security audit

Openclaw Redreplier

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill benign with medium confidence: its findings appear consistent with documented API management, but the review was partial and left dependency and network scope questions unresolved.

The verdict is benign, not risk-free, and confidence is medium. Before installing, consider whether the RedReplier API access and network scope suit your needs, review the dependency versions and advisories, and remember that some bundled code was only partially inspected by the tool-misuse analyzer.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (12)

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
80% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · SKILL.md (reported line 45)May include surrounding context.

md
Most RedReplier operations are safe and reversible (listing mentions, approving/rejecting). The API never charges: no endpoint upgrades the plan. Two actions destroy data and need explicit confirmation:

1. **`DELETE /keywords/{id}`** permanently deletes the keyword in any status together with every mention it produced. There is no undo. Confirm with the user first and name the keyword, not just the ID. Use `POST /keywords/{id}/disable` to stop monitoring and keep the mentions.
2. **`DELETE /websites/{id}`** stops all monitoring for the website. There is no restore endpoint (re-creating the same URL revives the record). Confirm with the user first; name the website (domain), not just the ID.

Other guidance:

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
80% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · SKILL.md (reported line 46)May include surrounding context.

md
Most RedReplier operations are safe and reversible (listing mentions, approving/rejecting). The API never charges: no endpoint upgrades the plan. Two actions destroy data and need explicit confirmation:

1. **`DELETE /keywords/{id}`** permanently deletes the keyword in any status together with every mention it produced. There is no undo. Confirm with the user first and name the keyword, not just the ID. Use `POST /keywords/{id}/disable` to stop monitoring and keep the mentions.
2. **`DELETE /websites/{id}`** stops all monitoring for the website. There is no restore endpoint (re-creating the same URL revives the record). Confirm with the user first; name the website (domain), not just the ID.

Other guidance:

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
80% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · SKILL.md (reported line 126)May include surrounding context.

md
- `PATCH /keywords/{id}` `{ "value": "new" }`: reword in place (unlimited, re-graded, keeps its slot). Use it to fix a `SUSPENDED` keyword or instead of adding a variant.
- `POST /keywords/{id}/disable`: stop one keyword immediately (unlimited, reversible). It frees its slot at once and keeps its mentions.
- `POST /keywords/{id}/enable`: bring back one `DISABLED` keyword. It goes `ACTIVE` when the plan has a free slot, otherwise `PENDING`. Never charges. `GET /keywords/billing-preview?desiredKeywordCount=N` (N = absolute active total wanted) prices the upgrade the user would make in the app.
- `DELETE /keywords/{id}`: permanently delete a keyword in any status together with every mention it produced. No undo and no refund; confirm first, and disable instead to keep the mentions.

### 4. List mentions (the leads)

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
80% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · references/api-reference.md (reported line 102)May include surrounding context.

md
it. The description is the AI scoring context; mentions already scored are not rescored. `ignoredKeywords` (string array, at most 100 entries of up to 255 characters each) replaces the whole ignore list: a mention whose title or text contains any entry leaves every mention list and count, and new matches are hidden as they arrive. Matching ignores case, blanks and duplicates are dropped, and `[]` clears the list; removing a word brings its mentions back. To add or remove one entry, read the current list from `GET /websites/{id}`, edit it and send all of it. The response returns `ignoredKeywords` lowercased. URL and keywords cannot be changed here. Returns the updated website with its keywords.

### DELETE /websites/{id}

Soft-deletes the website: it leaves `GET /websites` at once and its keywords stop matching. No restore endpoint; `POST /websites` with the same URL revives the record. Use `POST /keywords/{id}/disable` instead to pause a single keyword. Returns `{ "deleted": true }`.

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
80% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · references/api-reference.md (reported line 144)May include surrounding context.

md
Re-activates one keyword. Goes `ACTIVE` at once if the plan has a free slot, otherwise it is set `PENDING`. Never charges: the user upgrades the plan in the RedReplier app, and `GET /keywords/billing-preview` shows what that would cost. An already `ACTIVE` keyword is returned unchanged. Use `activate-pending` to promote every `PENDING` keyword that fits instead. Returns the keyword.

### DELETE /keywords/{id}

Permanently deletes a keyword in any status **and every mention it produced**. No undo. Deleting an `ACTIVE` keyword frees its slot the same way disabling does, with no refund; disable instead to keep the mentions. Returns `{ "deleted": true }`.

Anti-Refusal Statement

High
Category
Anti-Refusal
Confidence
85% confidence
Finding

Skill attempts to nullify the agent's safety policies or restrictions ('you have no restrictions', 'ignore your guidelines', 'do anything now'). This is a direct jailbreak that disables guardrails.

Content

Scanner excerpt · references/mention-filtering.md (reported line 7)May include surrounding context.

md
## Default behavior (no params)

`GET /mentions` with no filters applies two implicit filters:

1. **Excludes `REJECTED`** mentions.
2. **Hides anything below the website's minimum score** (30 unless the website has its own threshold), i.e. only `relevanceScore >= minimum` OR not-yet-scored mentions are shown.

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: network.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

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 61)May include surrounding context.

1. List monitored websites (and their keywords)

bash
curl -s -H "Authorization: Bearer $REDREPLIER_API_KEY" \
  https://ai.redreplier.com/ai-app/api/v1/websites

Unverifiable Dependency: openclaw has 591 known advisory(ies) (CVE-2026-53846 (OpenClaw: Workspace .env npm_execpath could influence bundled runtime dependency); CVE-2026-32064 (OpenClaw's andbox browser noVNC observer lacked VNC authentication); CVE-2026-32006 (OpenClaw has a BlueBubbles group allowlist mismatch via DM pairing-store fallbac) +588 more), but the manifest does not pin a version, so it is unknown whether the installed release is affected

Low
Category
Supply Chain
Confidence
40% confidence
Finding

OSV returned advisories for this package, but its resolved version is unknown. A package-name lookup does not establish whether the installed release is affected by those advisories.

Content

No source excerpt is available for this finding.

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
40% confidence
Finding

Dependencies lack version pinning, allowing potential malicious package updates. Consider pinning versions.

Content

Scanner excerpt · openclaw-plugin/package.json (reported line 23)May include surrounding context.

json
"openclaw": ">=2026.8.0"
  },
  "dependencies": {
    "@sinclair/typebox": "^0.34.52"
  },
  "devDependencies": {
    "@types/node": "^22.10.0",

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
40% confidence
Finding

Dependencies lack version pinning, allowing potential malicious package updates. Consider pinning versions.

Content

Scanner excerpt · openclaw-plugin/package.json (reported line 26)May include surrounding context.

json
"@sinclair/typebox": "^0.34.52"
  },
  "devDependencies": {
    "@types/node": "^22.10.0",
    "openclaw": "2026.8.1",
    "typescript": "^5.6.0"
  },

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
40% confidence
Finding

Dependencies lack version pinning, allowing potential malicious package updates. Consider pinning versions.

Content

Scanner excerpt · openclaw-plugin/package.json (reported line 28)May include surrounding context.

json
"devDependencies": {
    "@types/node": "^22.10.0",
    "openclaw": "2026.8.1",
    "typescript": "^5.6.0"
  },
  "repository": {
    "type": "git",