The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the local repository (such as package.json, compose.yaml, and .env files) to auto-detect the tech stack and suggest configurations.
Ingestion points: detect-stack.sh, scan-host-ports.mjs, and direct file reads in SKILL.md (Workflow Step 1).
Boundary markers: Not explicitly defined for the output of detection scripts.
Capability inventory: Execution of docker, docker compose, node, and local shell scripts across the project tree.
Sanitization: The detection scripts use regular expressions to extract specific version strings and port numbers, providing some structural validation of the data.
[COMMAND_EXECUTION]: The script scripts/port-check.sh uses the Bash /dev/tcp syntax to check for port availability.
Evidence: The script attempts to connect to 127.0.0.1 on specific ports using echo >/dev/tcp/127.0.0.1/"$port".
Analysis: While this pattern is often associated with reverse shells, in this context it is used strictly for local connectivity testing as a fallback when standard tools like nc or lsof are not present on the host system.

docker-local-dev