codex-tracker-suite

Pass

Audited by Gen Agent Trust Hub on Nov 1, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the subprocess module to execute the local codex command for session management. Evidence: scripts/codex_tracker.py uses subprocess.Popen to launch codex app-server.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests historical transcript data from previous user conversations, which creates a surface for indirect prompt injection if those transcripts contain instructions for the agent. 1. Ingestion points: scripts/codex_tracker.py reads JSON-formatted transcript data from the codex process's stdout. 2. Boundary markers: Absent; the skill does not currently implement specific delimiters or 'ignore' instructions for the ingested transcript text. 3. Capability inventory: The skill can execute the codex command and read local files in ~/.codex. 4. Sanitization: Data is printed to stdout or returned as JSON; basic character truncation is applied to long messages in the show command.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 1, 2026, 10:13 AM