The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses Bash tools in SKILL.md and several shell scripts (publish.sh, run-gates.sh, regenerate-pack.sh) for pipeline orchestration. These commands are restricted to local development tasks such as running tests, checking synchronization, and building models using local tools like npx, python3, and bash.
[EXTERNAL_DOWNLOADS]: The skill fetches resources from trusted platforms. It uses npx openpersona for persona creation and npx @agentplanet/acn for on-chain registration on the Base network. These target well-known package registries and established blockchain services.
[DYNAMIC_EXECUTION]: The scripts/memory.js and scripts/state-sync.js files use node to manage runtime state and memory. While they process JSON data, they follow safe practices for local file-based persistence without executing untrusted strings.
[DATA_EXFILTRATION]: The skill mentions ACN (Agent Communications Network) gateways and on-chain identity registration. These are documented features for agent interoperability and identity, using the vendor's designated gateway (acn-production.up.railway.app). No sensitive file exfiltration was detected.
[CREDENTIALS_SAFE]: The generated/persona-secondme-skill/soul/injection.md file explicitly instructs the agent to store credentials in the host's designated directory and warns against storing secrets in conversation memory or state files, following security best practices.

secondme-skill