The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses several shell commands to manage a local server and process job files.
Evidence: Uses curl for health checks, nohup for server persistence, open for user interaction, and various prompt-align CLI commands.
[EXTERNAL_DOWNLOADS]: Clones the prompt-align repository from GitHub if it is not found in the local development environment.
Evidence: git clone https://github.com/johnlindquist/prompt-align ~/prompt-align is defined in the Requirements section.
[REMOTE_CODE_EXECUTION]: Executes TypeScript code from the cloned repository using the Bun runtime.
Evidence: Commands such as bun <dir>/src/server.ts execute scripts that are downloaded and updated at runtime.
[PROMPT_INJECTION]: Vulnerable to indirect prompt injection as the agent's core behavior for task implementation is driven by external JSON content modified during the 'Edit' phase.
Ingestion points: Reads task instructions from ~/.prompt-align/jobs/<jobId>/result.json in Phase 3.
Boundary markers: Absent; the skill is instructed to treat the JSON content as a binding contract and standing authorization.
Capability inventory: Repository inspection, file modification, and command execution for verification.
Sanitization: No sanitization or verification of the human-steered or edited content in the JSON results is described.

align