The Agent Skills Directory

[PERSISTENCE]: The skill includes functionality to install systemd user services and timers (located in scripts/ar/schedule.py) at ~/.config/systemd/user/ to execute reporting tasks automatically at scheduled intervals.
[DATA_EXFILTRATION]: In scripts/ar/tokens.py, the skill scrapes local Claude and Codex transcript files (~/.claude/projects/ and ~/.codex/sessions/) to aggregate LLM token usage. While this data is used for metrics within the report, chat transcripts may contain sensitive information.
[INDIRECT_PROMPT_INJECTION]: The skill has a significant attack surface as it processes untrusted content from multiple external sources (Git commits, Plane board tickets, and Candystore events). Malicious data in these sources could potentially influence the behavior of the agent during the composition stage. The skill attempts to mitigate this through strict output linting and restricting the agent's tool permissions.
[DYNAMIC_EXECUTION]: The skill uses importlib.import_module in scripts/ar/cli.py for command routing and eval within scripts/run.sh to parse shell configuration derived from JSON data.

activity-report