The Agent Skills Directory

[DATA_EXFILTRATION]: The skill reads sensitive configuration files to verify credentials.
Evidence: The instructions direct the agent to "Cross-check .env for the relevant credentials" in step 1 of SKILL.md.
[COMMAND_EXECUTION]: The skill executes multiple local scripts and system commands to manage the agent service and database.
Evidence:
Uses launchctl and systemctl to check service status.
Runs scripts/q.ts via pnpm exec tsx to query the data/v2.db SQLite database.
Runs setup/index.ts via npx tsx for Telegram pairing.
Uses the ncl CLI tool to list groups and wirings.
Executes scripts/init-first-agent.ts to perform final wiring and initialization.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user input and interpolates it directly into shell command arguments, creating a vulnerability surface for command injection.
Ingestion points: USER_HANDLE, DISPLAY_NAME, and AGENT_NAME are collected as free-form text from the operator in SKILL.md step 2.
Boundary markers: Absent. The skill does not use delimiters or instructions to ignore embedded commands in these variables.
Capability inventory: The skill triggers a shell execution of scripts/init-first-agent.ts in step 5, passing the user-controlled variables as arguments.
Sanitization: Absent. There are no instructions to sanitize or validate the user input before it is passed to the shell script.

init-first-agent