The Agent Skills Directory

[DYNAMIC_CONTEXT_INJECTION]: The skill executes multiple shell commands and local binaries (e.g., ops-infra, ops-git, ops-unread) at load time using the !command syntax. It also runs a bash loop to read project paths from a local registry and extract progress data from markdown files.
[PERSISTENCE]: The skill implements a persistence mechanism by using the CronCreate tool to schedule recurring briefings. It prompts the user to establish daily or weekday execution schedules, allowing the agent to maintain active tasks over time.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources, introducing a vulnerability surface where malicious content could influence agent behavior.
Ingestion points: External data is ingested from WhatsApp messages, Gmail search results, and PR titles.
Boundary markers: The skill lacks explicit delimiters or instructions to ignore embedded commands within the processed data.
Capability inventory: The agent has access to Bash, TaskCreate, CronCreate, and WebFetch tools.
Sanitization: There is no evidence of input validation or sanitization for the unread messages and external project data before it is presented to the agent for analysis.
[EXTERNAL_DOWNLOADS]: The skill retrieves data from the official Google Calendar API via WebFetch as a fallback mechanism to enrich the briefing data.
[COMMAND_EXECUTION]: The skill frequently invokes shell commands and project-specific CLI tools (wacli, gog, jq) to aggregate infrastructure, code repository, and communication status.

ops-go