The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to manage and synthesize data from multiple sub-agents (workers), which creates a surface for indirect prompt injection.
Ingestion points: The agent is instructed to read and integrate worker reports, results, and material claims from independent worker scopes as defined in SKILL.md.
Boundary markers: The skill mitigates risks by requiring structured briefs for each worker and assigning disjoint write scopes to prevent accidental context bleed or collision.
Capability inventory: The skill utilizes the agent's file system and command execution capabilities to run a local initialization script and perform verification checks.
Sanitization: Robust instructions are provided requiring the agent to manually inspect all changed paths and verify every material claim against a rubric before integration into the main project.
[COMMAND_EXECUTION]: The skill instructs the agent to execute a local Python script to initialize the mission workspace.
Evidence: The SKILL.md file contains instructions to execute python3 "$SKILL_DIR/scripts/start_mission.py".
Context: This command is used for legitimate setup purposes, such as creating directories for handoffs and worker reports, and generating a mission ledger template.

dynamic-harness