Back to skill

Security audit

Public careers board

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed public job-board fetcher with constrained network access and user-directed local output, not a credential, payment, persistence, or destructive tool.

Install only if you are comfortable with a Node-based skill making live requests to the named public careers sites and optionally writing a caller-specified output file. Avoid running the bundled recipe/boards.mjs directly unless you intentionally want its separate multi-board evidence behavior.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (24)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 19)May include surrounding context.

md
reers board and needs the source and coverage stated. The installed command is `scripts/cli.mjs`. A description match does not select this skill, does not run t

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

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

md
reers board and needs the source and coverage stated. The installed command is `scripts/cli.mjs`. A description match does not select this skill, does not run t

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 87)May include surrounding context.

md
reers board and needs the source and coverage stated. The installed command is `scripts/cli.mjs`. A description match does not select this skill, does not run t

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 40)May include surrounding context.

md
node scripts/predicates.mjs observation.json

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 89)May include surrounding context.

md
node scripts/predicates.mjs observation.json

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 59)May include surrounding context.

md
`recipe/boards.mjs` is the upstream file. Do not run it as the installed command. Its own CLI writes an evidence file and reads every board. This command reads

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 82)May include surrounding context.

md
`recipe/boards.mjs` is the upstream file. Do not run it as the installed command. Its own CLI writes an evidence file and reads every board. This command reads

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 85)May include surrounding context.

md
`sources/magnite.mjs`

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 88)May include surrounding context.

md
`scripts/file-boundary.mjs`

Memory Manipulation

High
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill manipulates agent memory, state, or stored context. Memory corruption can alter personality, override safety rules, or cause unpredictable behavior.

Content

Scanner excerpt · scripts/file-boundary.mjs (reported line 19)May include surrounding context.

js
'A symlink parent fails that open and is refused rather than followed.',
  'The opened directory device and inode are compared with a fresh lstat of the same path.',
  'A mismatch is an ambiguous parent change and is refused rather than followed.',
  'The new name is created with O_CREAT|O_EXCL|O_NOFOLLOW on that directory descriptor, so the final component is not followed and an existing inode is not replaced.',
  'The directory descriptor, not a later path lookup, chooses where the name is created.',
  'This is not a universally race-proof check-then-open.',
  'This guarantee uses the Linux /proc/self/fd directory-descriptor route.',

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
91% confidence
Finding

The skill advertises executable behavior with network and shell capabilities but does not declare any explicit tool scope or permissions boundary in the manifest. That creates a governance and least-privilege gap: an agent or reviewer cannot tell from the metadata what external effects are intended, which increases the chance of overbroad execution or unsafe deployment.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The manifest says the skill should fetch current openings from one supported company careers board supplied to scripts/cli.mjs, but this code unconditionally fetches Acxiom, LiveRamp Ashby, and LiveRamp Workday together in buildLiveEvidence(). The CLI then writes the combined results to evidence/boards-live.json, which is broader behavior than a single-board reporting skill and adds persistent local output not described in the manifest.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · recipe/boards.mjs (reported line 30)May include surrounding context.

js
};
const ashbyCtx = {
  boardUrl: "https://jobs.ashbyhq.com/liveramp-inc",
  source: "https://api.ashbyhq.com/posting-api/job-board/liveramp-inc",
  fetchedAt: ctx.fetchedAt,
};
const good = { title: "One", externalPath: "/job/A/One_1", locationsText: "Remote" };

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · recipe/boards.public.test.mjs (reported line 12)May include surrounding context.

js
};
const ashbyCtx = {
  boardUrl: "https://jobs.ashbyhq.com/liveramp-inc",
  source: "https://api.ashbyhq.com/posting-api/job-board/liveramp-inc",
  fetchedAt: ctx.fetchedAt,
};
const good = { title: "One", externalPath: "/job/A/One_1", locationsText: "Remote" };

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · recipe/boards.public.test.mjs (reported line 22)May include surrounding context.

js
test("request gate rejects insecure protocols and embedded credentials", () => {
  for (const url of [
    "http://api.ashbyhq.com/posting-api/job-board/liveramp-inc",
    "https://user:secret@api.ashbyhq.com/posting-api/job-board/liveramp-inc",
    "https://api.ashbyhq.com.evil.example/posting-api/job-board/liveramp-inc",
  ]) assert.equal(gateRequest(url).ok, false);

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · recipe/boards.public.test.mjs (reported line 24)May include surrounding context.

js
for (const url of [
    "http://api.ashbyhq.com/posting-api/job-board/liveramp-inc",
    "https://user:secret@api.ashbyhq.com/posting-api/job-board/liveramp-inc",
    "https://api.ashbyhq.com.evil.example/posting-api/job-board/liveramp-inc",
  ]) assert.equal(gateRequest(url).ok, false);
  assert.equal(gateRequest(ashbyCtx.source).ok, true);
});

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

This file exposes write/delete primitives (createExclusiveChild, writeNewFileOutside, and cleanup via unlink) even though the skill is described as a read-only public careers-board fetcher. The helpers are somewhat constrained against overwriting and symlink traversal, but they still expand the skill's capability from data retrieval into local filesystem mutation, which increases attack surface and creates an opportunity for unintended file creation or deletion if reached by other code paths.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The code creates new files with exclusive write access and, on failure, deletes the created path, which are safety-relevant filesystem modifications. While the implementation is careful, there is no confirmation prompt, user-facing log/print, or inline user disclosure describing these write/delete effects in the code itself.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The code provides general-purpose file creation functionality that can write arbitrary bytes to caller-selected paths outside declared roots, which is unjustified for a job-board fetcher. Even with defensive checks, this capability is dangerous in context because it enables persistence or tampering on the local host if an attacker can influence arguments or invoke these helpers indirectly.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The manifest describes a skill for fetching current public job openings from supported careers boards, but this script instead validates pinned source files, creates a temporary copy, dynamically imports local code, and applies predicates to observation rows. This is not an obvious implementation detail of fetching a public careers board; it is a separate post-processing/execution path not reflected in the manifest description.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

For a skill whose stated purpose is to fetch and report public job openings, creating a scratch directory, writing executable module content, and dynamically importing it introduces code-execution behavior not justified by that purpose. Even though the source is integrity-pinned, this capability is materially different from simple board retrieval and reporting.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 31)May include surrounding context.

md
forbiddenPathPrefix: "/refreshFacet/",
  limit: 20,
  maxPages: 4,
  staleSuppliedSource: "https://api.smartrecruiters.com/v1/companies/Magnite/postings",
};

const ALLOWED_URLS = new Set([

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · sources/magnite.mjs (reported line 25)May include surrounding context.

js
forbiddenPathPrefix: "/refreshFacet/",
  limit: 20,
  maxPages: 4,
  staleSuppliedSource: "https://api.smartrecruiters.com/v1/companies/Magnite/postings",
};

const ALLOWED_URLS = new Set([

Description-Behavior Mismatch

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The recipe includes a call to use an unrelated third-party correspondence/task-help service that is outside the stated scope of fetching public careers boards. Even though it does not directly execute code or request secrets here, it introduces unnecessary off-platform redirection and could steer users to share operational details or data with an unvetted external service.

Content

No source excerpt is available for this finding.