The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources which constitutes a potential injection surface. 1. Ingestion points: The skill reads activity from Git commits, team chat messages (Slack, Discord, Teams, Google Chat), and issue trackers (GitHub, Linear, Jira) as described in Phase 1b and 3a of SKILL.md. 2. Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded commands within the gathered context. 3. Capability inventory: The skill uses shell tools (git) and network-capable MCP tools to post messages and create items. 4. Sanitization: While no automated sanitization is specified, the skill enforces a mandatory 'Approval Required' rule for all external write operations (Phase 2b, 3b, 3c), ensuring a human reviews all generated content before it is executed.
[SAFE]: The skill implements security best practices, such as instructing the agent to ensure the playbook configuration file is excluded from version control (.gitignore) to protect environment identifiers. It also explicitly includes a dedicated 'Security' classification for feedback triage, instructing the agent to flag concerns immediately rather than creating public issues.

team-update