The Agent Skills Directory

Indirect Prompt Injection (HIGH): The skill's core function is to ingest data from external platforms where content is controlled by third parties.
Ingestion points: Reads data from 'Code Review' (PRs/MRs), 'Issues' (Linear/Jira), and 'CI/CD' (error logs) as defined in the 'Data Sources' section of SKILL.md.
Boundary markers: No explicit boundary markers or 'ignore embedded instructions' warnings are defined for the aggregated output.
Capability inventory: The skill explicitly requires 'shell access' and executes local scripts (./scripts/sitrep.ts). The agent is instructed to use the report to 'Identify attention-needed items' and 'Plan work'.
Sanitization: No sanitization logic is mentioned for the text gathered from external titles, bodies, or comments.
Command Execution (MEDIUM): The skill relies on shell execution of a local TypeScript script and various CLI tools.
Evidence: The <scripts> section defines the execution of ./scripts/sitrep.ts with various flags.
Risk: If the script logic or the agent's interpretation of the script's output is compromised via indirect injection, the 'shell access' capability provides a high-impact exploitation path.
Data Exposure (LOW): Accesses sensitive repository and project metadata.
Evidence: Queries VCS state, PR decisions, and internal issue trackers.
Detail: While consistent with the skill's purpose, the aggregation of this data into a single report increases the impact of potential exfiltration if a prompt injection occurs.

status-reporting