Back to skill

Security audit

CORE CONSTITUTION MANIFEST API Spec v1.0.2

Security checks for vulnerabilities and agentic risk

Overview

This is a non-executable API specification, but it publishes a shared governance API key without clear scope, expiry, or safety limits.

Install only if you are comfortable with a documentation-only skill that contains a shared public API key. Before using the key, confirm it is sandboxed, low-privilege, rate-limited, and safe to redistribute; review or pin the separate conformance-test-suite package before installing it.

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)

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:95
Finding

Plaintext Shared Authentication Key Published in Skill Documentation

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:95
Vulnerability Type: Hardcoded shared credential
Risk Level: Medium

Vulnerable Code Snippet:

text
v19-e5d585e28439decc614f09f91c4caa8c

Technical Analysis

The document publishes a static value intended for use in the X-Governance-Key authentication header. Although it is presented as a public demonstration key, the audited file does not define an expiration time, endpoint restrictions, rate limits, environment isolation, or authorization scope.

A shared static credential cannot reliably identify individual callers and is available to anyone who can read or redistribute the Skill. If a deployed governance service accepts this value, possession of the Skill is sufficient to impersonate other users of the same credential. The API implementation was not included in the project, so whether the key remains active and what permissions it grants could not be verified.

Attack Path

  1. An attacker obtains SKILL.md from the Skill package.
  2. The attacker extracts the published key.
  3. The attacker identifies a deployed governance API that accepts X-Governance-Key.
  4. The attacker sends requests with the published value in that header.
  5. If the key is active, the service processes requests under the shared credential's permissions, without reliable per-user attribution.

Impact Assessment

The maximum impact depends on the server-side privileges assigned to the key. Potential impact includes unauthorized access to governance queries, constitution validation, bootstrap status, attention status, stability history, or audit records exposed to that credential.

If the key can access mutation-capable or sensitive endpoints, its publication could also permit unauthorized operations or disclosure of governance data. The repository does not contain the service implementation or authorization policy, so elevated privileges cannot be confirmed.

Remediation
View remediation

Remediation Suggestions

  • Remove the shared key from the distributed Skill and immediately revoke or rotate it if it has ever been accepted by a deployed service.
  • Issue unique, short-lived credentials through an authenticated enrollment process.
  • Store credentials in an approved secret manager or runtime environment variable rather than documentation or source-controlled files.
  • Restrict demonstration credentials to an isolated sandbox with no production data or mutation privileges.
  • Apply explicit endpoint scopes, expiration, rate limits, quotas, and source restrictions.
  • Record activity against unique identities rather than a shared principal.
  • Add automated secret scanning and credential-revocation procedures to the publication workflow.

T08 · Insecure Dependencies

Warning
Location
SKILL.md:115
Finding

Unpinned Third-Party Package Installation

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:115
Vulnerability Type: Unverified and mutable third-party dependency
Risk Level: Medium

Vulnerable Code Snippet:

bash
clawhub install v19-conformance-test-suite

Technical Analysis

The installation instruction refers to an external package solely by name. It does not pin an immutable version or cryptographic digest, identify a verified publisher, or provide source code within the audited project.

Consequently, the code installed by this command can differ from the content reviewed here. Registry compromise, publisher-account compromise, package-name takeover, or a malicious future release could cause users to retrieve unsafe content. Whether installation executes lifecycle hooks or grants the package additional privileges depends on ClawHub's behavior and was not verifiable from the sole audited file.

Attack Path

  1. An attacker compromises the package publisher, distribution registry, or package namespace.
  2. The attacker publishes a malicious release under v19-conformance-test-suite.
  3. A user follows the documented installation command without an immutable version or digest.
  4. The package manager resolves the attacker-controlled release.
  5. Malicious content is installed and may execute if the package manager invokes installation hooks or when the installed package is subsequently used.

Impact Assessment

The potential scope is bounded by the privileges of the user or Agent running the installation and the package. Depending on package-manager behavior, compromise could expose local files, credentials, Agent data, or network-accessible resources and could permit arbitrary actions available to that process.

No malicious package content or actual execution is present in the audited repository. This finding concerns the unsafe, mutable dependency acquisition instruction and its resulting supply-chain exposure.

Remediation
View remediation

Remediation Suggestions

  • Pin the package to a reviewed immutable version and cryptographic digest.
  • Document and verify the expected publisher identity and authoritative source repository.
  • Require signed packages and verify signatures before installation.
  • Review the resolved package source, manifest, lifecycle hooks, and transitive dependencies.
  • Use a lockfile or equivalent integrity manifest where supported.
  • Install and test the package in a sandbox with minimal filesystem, credential, and network access.
  • Configure automated monitoring for publisher changes, unexpected package updates, and integrity mismatches.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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)

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The document directly embeds a usable API key and labels it as a public experience key, which creates an immediate credential exposure risk. Even if intended for demo use, publishing a live token in an API specification invites unauthorized use, quota exhaustion, abuse of the backing service, and possible pivoting if the key has broader permissions than documented.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The exposed key is presented without any warning, scope limitation, expiration note, or safety guidance, increasing the likelihood that readers will treat it as a normal production credential and reuse it broadly. In the context of a governance API, unauthorized access could expose internal status data, enable abusive registration or query activity, and undermine trust in the control plane.

Content

No source excerpt is available for this finding.