Kandler3 Yandex Tracker
Use yandex_tracker_client to interact with Yandex Tracker API v2. Use the execution, secret-management, and temporary-file facilities available in the current runtime.
Workflow
- Identify whether the request is read-only or mutating. Do not create, update, transition, delete, or bulk-change data unless the user has authorized that action and its scope is clear.
- Read only the topic references needed for the request.
- Before the first API call in an environment, follow setup and authentication.
- For multi-step work, create one self-contained Python script in a runtime-appropriate temporary or working directory. Combine related queries and mutations in that script, then print a concise structured result.
- Materialize lazy API iterables with
list(...)before counting, sorting, reusing, or summarizing them. - Discover queue-specific values instead of guessing custom field keys, transition IDs, resolutions, users, or sprint IDs.
- After a mutation, report the affected issue keys and the confirmed result. For asynchronous bulk changes, wait for completion and surface failures.
Safety
- Treat tokens and organization identifiers as secrets. Read them from the runtime's secret or environment mechanism; never print, persist, or embed them in generated scripts.
- Use least-privilege credentials. Do not broaden permissions or install dependencies without the authorization required by the current runtime.
- Inspect the current issue state before applying a context-dependent update. For list-valued fields, distinguish full replacement from
add/removemutations. - Before a broad or destructive bulk operation, show or otherwise verify the exact issue set and intended change.
Topic index
- Environment, credentials, client initialization, and portable execution: read setup and authentication before the first API call or when setup fails.
- Queries, filters, custom fields, pagination, and aggregation: read search and reporting for discovery or reporting tasks.
- Get, create, update, and transition issues: read issue lifecycle for issue mutations and status changes.
- Comments, mentions, attachments, and issue links: read collaboration only when the request involves those resources.
- Worklogs, queues, users, boards, and sprints: read worklogs and planning for time tracking or planning metadata.
- Multi-issue updates, transitions, and moves: read bulk operations before any bulk change.
- Dynamic objects, field shapes, and exceptions: read object reference when inspecting unfamiliar return values or handling API errors.
- OpenClaw-specific installation and configuration: read OpenClaw compatibility only when the skill runs in OpenClaw.