The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the local project environment to generate prose updates.
Ingestion points: SKILL.md (Workflow Step 1) specifies reading CHANGELOG.md, commit history, and release records from the target project.
Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded commands within the project data it reads.
Capability inventory: The skill utilizes scripts/profile_store.py for reading/writing configuration files and has access to read the target project's filesystem via host tools.
Sanitization: No explicit sanitization or filtering of the ingested changelog or commit data is documented before it is processed by the agent.
[COMMAND_EXECUTION]: The skill includes local Python scripts to manage state and validate its own structure.
scripts/profile_store.py: Manages reading and writing user preferences and skill-specific records to a JSON profile file (e.g., in ~/.config/agent-skills/). It includes safeguards such as blocking sensitive keys (passwords, tokens) and setting restrictive file permissions (0600).
scripts/validate_skill.py: A validation utility used to check the integrity of the skill source, which requires the PyYAML package.

lov-daily-post