The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes local TypeScript utility scripts to perform environment checks and data transformation. scripts/detect_runner.ts probes the system for available JavaScript runtimes using spawnSync on a hardcoded list of candidates. scripts/intake_issue.ts and scripts/linear_metadata.ts are used to parse user input and manage Linear metadata snapshots. These executions are constrained to the skill's own scripts and follow safe implementation patterns.
[PROMPT_INJECTION]: The skill acts as an ingestion point for untrusted user data by transforming rough reports into structured Linear issues.
Ingestion points: Raw bug reports and feature ideas are ingested through user prompts and structured YAML files processed by scripts/intake_issue.ts.
Boundary markers: The workflow uses markdown templates and specific field requirements to structure the output, although it lacks explicit boundary delimiters or 'ignore embedded instructions' markers for user-provided text in the final rendered body.
Capability inventory: The agent can write to Linear issues via MCP tools (save_issue) and execute local scripts via shell commands.
Sanitization: The helper scripts validate that the input is a valid YAML mapping and contains required fields, but do not perform content-level filtering or sanitization of the raw text for prompt injection payloads.

linear-create-issue