The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from an external source (Linear issues), which could contain malicious instructions designed to influence the analysis agent.
Ingestion points: The catalyst-dev:linear-research agent fetches backlog issues in JSON format in Step 1 of the grooming process (SKILL.md).
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are present in the backlog-analyzer prompt.
Capability inventory: The skill has access to Task, Read, and Write tools, and spawns other agents to perform research and analysis.
Sanitization: No sanitization of issue titles or descriptions is performed before the data is passed to the analysis agent.
[DYNAMIC_EXECUTION]: The skill dynamically generates a bash script at runtime based on recommendations from the analysis agent.
Evidence: The skill assembles an update script in thoughts/shared/pm/reports/, which is then granted execution permissions via chmod +x (SKILL.md, Step 5).
[COMMAND_EXECUTION]: The skill relies on executing local shell scripts and third-party CLI tools to manage project context and fetch data.
Evidence: Executes ./scripts/validate-thoughts-setup.sh, ./scripts/check-prerequisites.sh, and linearis CLI commands throughout the workflow.

groom-backlog