The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes Salesforce CLI (sf) commands using the Bash tool to manage org lifecycles. These operations include creating, listing, displaying, and deleting scratch orgs, which are all within the stated scope of the skill.
[SAFE]: The skill contains explicit security guardrails to protect credentials. It instructs the agent to never run commands like sf org display --verbose or sf org open --url-only because they return live session tokens. It also ensures that any JSON artifacts written to the filesystem preserve the CLI's built-in redaction of sensitive fields like accessToken.
[SAFE]: The utility script assets/derive-alias.sh is used for deterministic alias generation. It processes local CLI output using standard tools (tr, sed, jq) without performing network operations or accessing sensitive user files outside of the Salesforce CLI context.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from project configuration files (sfdx-project.json, config/*.json), which represents a theoretical attack surface. However, the risk is minimal as the data is passed to the official Salesforce CLI for processing.
Ingestion points: Reads local Salesforce project configuration files.
Boundary markers: Absent for file reading, but instructions require the agent to relay CLI output verbatim.
Capability inventory: Executes Salesforce CLI commands and writes JSON artifacts to the workspace.
Sanitization: Explicitly redacts authentication tokens and restricts the use of flags that expose secrets.

dx-org-manage