The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the local project environment, including git commit history, README files, and project manifests, which could be manipulated to include instructions for the agent.
Ingestion points: Data is gathered from git log, README.md, and manifest files (like package.json or Cargo.toml) as described in the signal gathering phase of SKILL.md.
Boundary markers: The skill design includes a mandatory human-in-the-loop checkpoint, requiring the agent to show a diff-style preview and wait for user confirmation before executing any file edits.
Capability inventory: The skill possesses the capability to read project files and write to a central portfolio Markdown archive using file system tools.
Sanitization: The skill does not implement automated sanitization or escaping for the ingested signals, relying entirely on the user to identify anomalies during the preview phase.

portfolio-sync