The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to utilize local command-line tools for runtime management, specifically python3 <agentit-root>/router/runtime_cli.py for initializing and checking execution loops.
[INDIRECT_PROMPT_INJECTION]: The skill implements a 'Worker Context Contract' that projects project instructions and user-supplied constraints into delegated worker agents. This creates a potential surface for indirect prompt injection from processed data, though the skill mandates 'mechanical projection/validation' via router/worker_context.py and provides explicit warnings that context envelopes do not constitute host sandboxing.
Ingestion points: Worker Context Contract (SKILL.md)
Boundary markers: Use of 'bounded projection' and 'least-privilege capability envelope'
Capability inventory: Shell command execution via runtime_cli.py (SKILL.md)
Sanitization: Validation logic referenced in router/worker_context.py and router/runtime_cli.py
[DYNAMIC_EXECUTION]: The skill defines 'Loop Engineering' and 'Graph Engineering' systems where executable units are managed through deterministic contracts with observable outcomes, verifiers, and attempt budgets. These units are materialized as state files and DAGs within the .agentit/runtime/ directory.

architect-orchestrator