The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The bootstrap.sh and bootstrap.ps1 scripts download and execute installation scripts directly from the vendor's GitHub repository (github.com/ArchAstro/archdev). This pattern allows for the execution of remote code during the skill's setup phase.
[EXTERNAL_DOWNLOADS]: The skill fetches configuration and installation scripts from external URLs (specifically raw.githubusercontent.com) during the bootstrap and setup procedures.
[DYNAMIC_EXECUTION]: The skill is designed to execute pipeline steps defined in archdev.json configuration files. This allows for the execution of arbitrary shell commands as specified in the repository's policy and manifests.
[COMMAND_EXECUTION]: The skill instructs the agent to use the archdev CLI to perform system-level operations, including daemon management, repository registration, and running durable jobs.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted repository data to configure and execute automation, creating a surface for indirect prompt injection.
Ingestion points: The skill reads tracked manifests, CI workflow steps, test/build scripts, and existing .archdev definitions from the repository being automated (references/configuration.md).
Boundary markers: No specific prompt delimiters or "ignore embedded instructions" warnings are documented for the ingestion of this data.
Capability inventory: The skill can execute arbitrary commands via jobs run and jobs repo submit, and it can mutate GitHub Pull Requests through its daemon-based remediation watcher.
Sanitization: The skill mentions a branch-validation phase which includes an "adversarial review" to inspect findings and proposed fixes before execution.

jobs