The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute several local Python scripts to manage session state, including create_handoff.py, validate_handoff.py, list_handoffs.py, and check_staleness.py. These scripts are part of the skill's infrastructure and their execution is central to the intended workflow of session management.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill aggregates project metadata, including git branches, recent commits, and modified file lists, into handoff documents. To mitigate the risk of capturing sensitive data, the skill includes a validate_handoff.py script that specifically checks for the presence of API keys, passwords, and tokens before the handoff is finalized.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through the ingestion of external data files during the RESUME workflow.
Ingestion points: The agent is instructed to read generated markdown handoff files in Step 3 of the RESUME Workflow.
Boundary markers: Absent. The instructions do not specify using delimiters or warnings to ignore instructions potentially embedded within the handoff text.
Capability inventory: The agent uses the content of these files to determine 'Immediate Next Steps' and 'Current State', which influences its future autonomous actions.
Sanitization: The validate_handoff.py script checks for credentials and quality but does not appear to sanitize the text for malicious prompt instructions.

session-handoff