Back to skill

Security audit

One missed deadline ends a career. caseclaw makes sure it never happens. Point it at a matter and it auto-calculates every filing deadline from 600+ court rule sets. Throw a new client name at it and it checks conflicts against your entire case history in under a second. Start typing and it's already timing your work in six-minute increments. No browser tabs. No clunky SaaS dashboards. No $500/seat enterprise contracts. Just one command and the peace of mind that nothing is slipping through.

Security checks for vulnerabilities and agentic risk

Overview

This looks like a coherent legal case-management skill, but it installs an unaudited third-party CLI that may process confidential legal data and send some data to external services.

Review the `caseclaw` Homebrew tap, source, release provenance, and privacy/security terms before installing. Use an isolated environment or test data until you trust the binary, and treat Pro/Firm features plus email, Slack, and calendar alerts as external disclosure paths for client and matter metadata.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (1)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:5
Finding
Unpinned Executable Installation from a Mutable Third-Party Homebrew Tap## Vulnerability Details **File Location**: `SKILL.md`, line 5 **Vulnerability Type**: `T08: Insecure Dependencies` **Risk Level**: Medium **Complete Code Snippet**: ```yaml metadata: {"clawdbot":{"emoji":"🔱","requires":{"bins":["caseclaw"]},"install":[{"id":"brew","kind":"brew","formula":"legal-tools/tap/caseclaw","bins":["caseclaw"],"label":"Install caseclaw (brew)"}]}} ``` ### Technical Analysis The skill instructs the host environment to install the `caseclaw` executable from the external Homebrew formula `legal-tools/tap/caseclaw`. The dependency is referenced through a mutable package name without an immutable source commit, fixed package version, checksum, or cryptographic signature recorded in the audited project. Neither the Homebrew formula nor the executable's source is included in the project, which contains only `SKILL.md` and `_meta.json`. Therefore, the behavior and integrity of the code ultimately executed as `caseclaw` cannot be verified from this artifact. A future update to the tap, compromise of its publishing account or repository, or compromise of an upstream release could change the installed payload after this skill has been reviewed. This is a supply-chain weakness rather than evidence that the current external package is malicious. ### Attack Path 1. An attacker compromises the `legal-tools/tap` repository, its publishing credentials, the referenced formula, or an upstream artifact used by that formula. 2. The attacker changes the formula or upstream release so that it installs a malicious `caseclaw` executable. 3. A user loads or installs the skill in an environment where `caseclaw` is not already available. 4. The skill's installation metadata causes Homebrew to resolve the mutable `legal-tools/tap/caseclaw` formula. 5. Because this project specifies no trusted version or expected digest, the modified package may be accepted and installed without detection by the skill. 6. The user later in ...[truncated 1120 chars]
Remediation
## Remediation Suggestions 1. Pin the dependency to a reviewed, immutable release rather than relying solely on a mutable Homebrew formula name. 2. Record and verify a SHA-256 digest or a trusted cryptographic signature for the downloaded artifact before installation. 3. Pin the Homebrew tap or formula source to an immutable commit and ensure that the formula itself verifies the integrity of its upstream archive. 4. Include the formula and executable source in the auditable package, or provide reproducible-build documentation that allows reviewers to map the installed binary to reviewed source. 5. Protect release and tap publishing workflows with multi-factor authentication, least-privilege tokens, protected branches, mandatory review, signed tags, and provenance attestations. 6. Fail closed if signature, checksum, provenance, package version, or source verification does not match the approved release. 7. Display the exact package version and source to the user and require explicit confirmation before installing an executable that will process confidential legal data. 8. Run the executable with least privilege and restrict its filesystem and network access where practical. 9. Independently review the external formula and executable before treating the package's documented security, storage, or network-processing claims as verified.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Rogue AgentSelf-Modification, Session Persistence
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Missing User Warnings

Medium
Confidence
92% confidence
Finding
The skill prominently markets local-first handling, but the alerting and Pro/Firm API-backed features introduce transmission of sensitive legal matter data to email, Slack, calendar providers, and remote processing without an equally prominent warning at the point those commands are introduced. In a legal context, even metadata such as client names, matter titles, deadlines, and court information can be highly sensitive, so users may disclose protected information without informed consent.

Session Persistence

Medium
Category
Rogue Agent
Content
# 1. Run conflict check
caseclaw conflict check "New Client LLC" "Opposing Party Inc"

# 2. Create matter
caseclaw matter new --name "NewClient v. Opposing" --type litigation --court SDNY

# 3. Auto-generate deadlines
Confidence
60% confidence
Finding
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.

Static analysis

No suspicious patterns detected.