The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the use of shell commands such as mkdir -p to initialize documentation directories and grep to scan for existing record IDs within the project.
[PRIVILEGE_ESCALATION]: The instructions include using chmod +x to grant execution permissions to a newly created project-local shell script (.claude/hooks/decision-log-capture.sh). This is a standard requirement for platform hooks and does not involve system-level privilege changes.
[DYNAMIC_EXECUTION]: The skill generates a bash script and configures a platform-specific hook (PostToolUse) in .claude/settings.json to execute this script automatically. The script itself is a static template that outputs JSON context to the agent.
[INDIRECT_PROMPT_INJECTION]: The skill processes project data and conversation history to generate ADR files, which represents a theoretical attack surface for indirect injection if the source plans are malicious.
Ingestion points: Content from the agent's "plan mode" is used to populate the ADR template via the hook's additionalContext.
Boundary markers: None present; the skill relies on the agent's ability to follow the markdown template structure.
Capability inventory: The skill utilizes file writing, directory creation, and file reading (grep) to manage the decision log.
Sanitization: No explicit sanitization is performed on the plan content before it is transformed into a markdown record.

captains-log