The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute repository-specific setup commands, including npm install, cargo build, pip install -r requirements.txt, go mod download, mvn compile, ./gradlew compileJava, and dotnet restore based on the detected project type.
[DYNAMIC_EXECUTION]: The skill executes a project-local Python script located at ./.recursive/scripts/recursive-router-probe.py to refresh local discovery state and determine routing policies.
[INDIRECT_PROMPT_INJECTION]: The agent's routing decisions and setup behavior are governed by the contents of local configuration files (.recursive/config/recursive-router.json and recursive-router-discovered.json). Malicious modifications to these files could influence the agent's subsequent actions or model selection.
Ingestion points: Reads routing policy and discovery inventory from /.recursive/config/recursive-router.json and /.recursive/config/recursive-router-discovered.json (SKILL.md).
Boundary markers: None; the agent is instructed to "honor the routed policy" immediately after reading the files.
Capability inventory: Git worktree management, package installation via multiple package managers, and Python script execution.
Sanitization: No evidence of schema validation or sanitization is provided for the JSON configuration files before they are used to determine routing logic.

recursive-worktree