pm-skills
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill integrates with Atlassian's official Remote MCP service for Jira and Confluence access. It also references established industry frameworks from organizations such as Anthropic, Linear, and Scrum.org. These external references are appropriate for the skill's project management domain and follow security best practices.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external Jira issue data through
scripts/jira_snapshot_bridge.py, which constitutes a surface for indirect prompt injection if the project data contains malicious instructions designed to influence the agent's behavior. - Ingestion points: External Jira issue JSON is ingested via the
--inputflag inscripts/jira_snapshot_bridge.py. - Boundary markers: The skill uses strict JSON schema validation for input snapshots and enforces deterministic processing through Python scripts rather than direct prompt interpolation.
- Capability inventory: The skill can execute local Python scripts, perform Jira searches, and modify Jira issue statuses (gated by human review).
- Sanitization: The skill mitigates risks by converting external data into structured metrics (e.g., WIP, throughput, cycle time) and implementing an accountability gate (
scripts/delivery_loop_gate.py) that requires human approval for critical actions.
Audit Metadata