The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from external sources including Slack messages, GitHub pull request comments, and Linear issue updates.
Ingestion points: Data is retrieved via MCP tools from Slack (messages and threads), GitHub (PR and issue comments), and Linear (issue descriptions and comments) as described in references/tool-queries.md.
Boundary markers: The instructions do not specify the use of delimiters or specific boundary markers when interpolating untrusted content into the agent's context.
Capability inventory: The skill utilizes data retrieval tools across Slack, GitHub, Linear, and Google Calendar, and may utilize local git commands for repository history.
Sanitization: The skill provides strong instructions to redact sensitive internal URLs, cloud console links, Secrets Manager links, and credential-related information from the final standup draft.
[DATA_EXPOSURE_AND_EXFILTRATION]: While the skill accesses sensitive work activity, it implements proactive safeguards for data handling.
The instructions explicitly forbid the inclusion of raw private links, cloud console links, or credential values in the output.
Security-conscious handling of Slack bot tokens is mandated, requiring the use of environment variables or stdin and prohibiting the printing or persistence of tokens.

standup