The Agent Skills Directory

Indirect Prompt Injection Surface: The skill fetches external content such as mailing list bodies and GitHub comments to perform vote tallies. This content is untrusted and could potentially contain malicious instructions. The skill includes strong mitigations by explicitly instructing the agent to treat this content as data rather than instructions and to report any detected injection attempts in the injection_summary field.
Ingestion points: Vote thread bodies from mail archives, GitHub Discussion replies, and PR review comments.
Boundary markers: The skill provides clear instructions to treat external content as data and to ignore any embedded directives.
Capability inventory: The skill executes shell commands using gh (GitHub CLI) and python3, and performs network read operations to fetch vote data.
Sanitization: The agent is instructed to manually flag and ignore injection attempts, though no automated sanitization script is invoked on the data itself.
Command Execution: The skill executes local Python modules and CLI tools to perform its logic.
Evidence: Use of python3 -m setup_preflight for project verification and gh api for fetching GitHub discussion data.
External Data Fetching: The skill retrieves information from external services to aggregate vote signals.
Evidence: Connection to Apache mailing list archives (via PonyMail templates) and GitHub APIs to read thread contents.

vote-tally