Back to skill

Security audit

agent-daily-journal

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious with medium confidence because a scanner found a configurable remote endpoint may receive historical activity content despite local-only documentation; the scan was partial.

Review how the skill handles journal and activity data, especially whether its configurable model endpoint could be remote and what information would be sent. The scan covered all listed components but was marked partial because several referenced paths were missing from the bundle, so the evidence is incomplete. Consider installing only if the endpoint behavior and required tool capabilities fit your privacy and security expectations.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (6)

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

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.

The style template requires journal output in English without explicit user opt-in.

Medium
Category
Output Handling
Confidence
95% confidence
Finding

The Language field explicitly sets English, and lines 3–5 establish this file as the voice authority read on every run. Although the heading identifies an English template, the file does not explicitly offer a language choice or justify a fixed-language requirement.

Content

No source excerpt is available for this finding.

The documented local-only pipeline can send activity content to a remote model endpoint.

Medium
Category
Excessive Agency
Confidence
97% confidence
Finding

The endpoint is configurable through configuration or JOURNAL_OLLAMA_URL without a loopback restriction, and embedding and compression requests transmit historical user prompts and activity details to that endpoint. Although the default is loopback and transmission requires suitable model detection, a remote endpoint can receive sensitive history without explicit remote-processing consent, contradicting the local-only assurance.

Content

No source excerpt is available for this finding.

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · tests/test_journal_lint.py (reported line 127)May include surrounding context.

python
import subprocess

        script = ROOT / "scripts" / "journal_lint.py"
        proc = subprocess.run(
            [sys.executable, str(script), "--digest", str(FIXTURES / "digest_2026-03-14.md"),
             "--entry", str(FIXTURES / "entry_thin_2026-03-14.md"), "--no-semantic", "--json"],
            capture_output=True, text=True, encoding="utf-8",

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · tests/test_journal_lint.py (reported line 141)May include surrounding context.

python
import subprocess

        script = ROOT / "scripts" / "journal_lint.py"
        proc = subprocess.run(
            [sys.executable, str(script), "--digest", str(FIXTURES / "digest_2026-03-14.md"),
             "--entry", str(FIXTURES / "entry_good_2026-03-14.md"), "--no-semantic"],
            capture_output=True, text=True, encoding="utf-8",

Dynamic attribute access via getattr()

Low
Category
Dangerous Code Execution
Confidence
50% confidence
Finding

Dynamic getattr() with a non-literal attribute name can access arbitrary object attributes, potentially bypassing access controls.

Content

Scanner excerpt · scripts/journal_collect.py (reported line 520)May include surrounding context.

python
cfg = load_config(args.config)
    for attr, key in [("journal", "journal"), ("state_db", "state_db"), ("workspace", "workspace"),
                      ("day_start_hour", "day_start_hour"), ("max_windows", "max_windows")]:
        value = getattr(args, attr)
        if value is not None:
            cfg[key] = Path(value) if key in {"journal", "state_db", "workspace"} else value
    if not cfg.get("workspace"):