The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from Yandex Tracker issues, including summaries, descriptions, and comments, which could be used to host malicious instructions intended to influence the agent's behavior.
Ingestion points: Data is ingested via client.issues.find(), client.issues.get_all(), and other fetch methods described in references/search-and-reporting.md and references/collaboration.md.
Boundary markers: There are no instructions to use delimiters or explicit boundary markers when processing or summarizing external issue data.
Capability inventory: The agent has capabilities for network operations (via the Tracker API), file system writes (generating scripts and downloading attachments), and issue modification.
Sanitization: No specific sanitization or filtering of external issue content is mentioned in the instructions.
[DYNAMIC_EXECUTION]: The skill workflow involves generating and executing Python scripts to perform multi-step operations.
Evidence: SKILL.md (Workflow step 4) and references/setup-and-auth.md (Portable execution section) instruct the agent to create self-contained Python scripts in temporary directories to combine API queries and mutations.
[EXTERNAL_DOWNLOADS]: The skill requires an external Python dependency for its core functionality.
Evidence: references/setup-and-auth.md and references/openclaw.md mention the installation of the yandex_tracker_client package via pip.

yandex-tracker