The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from multiple external sources, creating an attack surface where embedded instructions could influence the agent's summary or subsequent actions.
Ingestion points: Slack message threads, GitHub PR titles/comments, Linear issue descriptions, and local meeting notes referenced in SKILL.md Phase 1.
Boundary markers: Absent. The instructions for subagents do not include specific delimiters or warnings to ignore malicious instructions within the processed data.
Capability inventory: Shell command execution via the gh CLI, network operations through Slack, Linear, and PostHog MCP tools, and file system write access for saving synthesis reports described in SKILL.md Phase 2.
Sanitization: Absent. The skill summarizes and cross-references external text without explicit sanitization or validation of the content's safety.
[COMMAND_EXECUTION]: The skill executes shell commands using the GitHub CLI to retrieve repository metrics.
Evidence: SKILL.md Phase 1 Agent 1 uses gh pr list and gh api to collect pull request data and contributor statistics.

comprehensive-analysis