The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a project-local script located at .specify/scripts/bash/check-prerequisites.sh with the --json flag to retrieve the absolute path of the active feature directory.
[COMMAND_EXECUTION]: The skill invokes git log using user-provided parameters from $ARGUMENTS to bound the timeframe of the session summary.
[PROMPT_INJECTION]: The skill processes untrusted data from external sources, including git log output, the conversation buffer, and local project files (spec.md, plan.md, tasks.md), which constitutes an indirect prompt injection surface.
Ingestion points: git log commits, project Markdown files, and the session conversation history.
Boundary markers: The instructions include explicit directives to avoid transcripts and long quotes, effectively delimiting processed data.
Capability inventory: The skill can write files to the FEATURE_DIR/sessions/ directory and execute the local check-prerequisites.sh script.
Sanitization: The agent is strictly instructed to summarize findings into a narrative format rather than reproducing raw content or diffs, which serves as a natural barrier against malicious instructions embedded in the source data.

speckit-opsmill-summary