funding-allocation

Pass

Audited by Gen Agent Trust Hub on Mar 23, 2026

Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it is instructed to 'Read the actual codebase' and process user-supplied '$ARGUMENTS' without implementing boundary markers or instructions to ignore embedded commands. An attacker could place malicious instructions within the analyzed codebase to influence the agent's behavior during the analysis phases.
  • [DATA_EXFILTRATION]: The 'Self-Evolution Telemetry' section attempts to access and modify files within ~/.claude/projects/. This involves interacting with hidden directories containing agent-specific metadata and project history. While no network exfiltration is observed, writing to internal configuration paths of the agent's execution environment is a violation of the principle of least privilege for a task focused on financial analysis.
  • [PROMPT_INJECTION]: The skill provides an evidence chain for indirect prompt injection vulnerability: (1) Ingestion points: The skill reads the 'actual codebase' and '$ARGUMENTS' (SKILL.md). (2) Boundary markers: Absent. (3) Capability inventory: The skill has file-write capabilities in the docs/ and ~/.claude/projects/ directories. (4) Sanitization: Absent.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 23, 2026, 10:56 AM
Security Audit — agent-trust-hub — funding-allocation