The Agent Skills Directory

[COMMAND_EXECUTION]: The skill modifies configuration files for external tools (.claude/settings.json and config.toml) to insert hooks that execute shell scripts (session_end_log.sh) automatically at the end of AI sessions or turns.
[COMMAND_EXECUTION]: Utilizes shell eval within codex_notify_wrapper.sh and session_end_log.sh to dynamically load environment variables exported by a Python helper script. While variables are quoted and keys are validated, this pattern involves executing shell code generated at runtime.
[PROMPT_INJECTION]: The skill processes the entire content of an Obsidian vault to generate Weekly Reports and Knowledge Maps. Malicious instructions embedded in vault markdown files could potentially influence the AI's summary or report generation.
Ingestion points: scripts/lint_vault.py and scripts/gen_vault_map.py read all .md files in the specified vault.
Boundary markers: No explicit delimiters or "ignore embedded instructions" warnings are used when interpolating file data into the report generation prompt.
Capability inventory: The skill has permissions to write to the file system (creating reports and maps), modify tool configurations, and execute shell scripts via hooks.
Sanitization: Shell arguments are sanitized using shlex.quote in maintain_env.py, and filenames are managed through Python's os module.

soia-pkm-maintain