Loading...
Loading...
Found 5 Skills
Builds and deploys data processing and ML training pipelines using TrueFoundry Workflows (built on Flyte). Use when creating DAGs, orchestrating multi-step tasks, scheduling ETL pipelines, or running ML training workflows.
Breaks natural-language problem descriptions into sub-tasks suitable for DAG nodes. The entry point of the meta-DAG. Identifies phases, dependencies, parallelization opportunities, and vague/pluripotent nodes that can't yet be specified. Uses domain meta-skills when available. Activate on "decompose task", "break down problem", "plan workflow", "what are the steps", "sub-tasks", "task breakdown". NOT for executing the decomposed tasks (use dag-runtime), building the DAG structure (use dag-planner), or matching skills to nodes (use dag-skills-matcher).
Dynamic orchestration engine that plans multi-step agent work as DAGs with Mermaid visualization.
Multi-agent coordination, inter-terminal messaging bus, distributed file/mutex locking with monotonic fencing tokens, orchestrating multi-stage DAG task pipelines, cluster health watchdog sweeping, and worker queues over Redis. Use when coordinating work between multiple AI assistants or terminal sessions, acquiring distributed mutex locks before editing shared files or running migrations/deployments, generating monotonic fencing tokens to prevent zombie writes, dispatching tasks or RPC queries to peer agents, producing/consuming from competing-consumer work queues, orchestrating multi-stage pipelines with automatic dependency resolution, reliably claiming tasks with leases and ack/DLQ handling, scattering tasks to a pool for quorum aggregation, sharing scratchpad memory, managing floor control in roundtable brainstorming, setting and checking run cancellation tokens, running blind consensus ballots without anchoring bias, electing resilient mesh leaders with automated lease failover, auditing cluster health and sweeping dead agent/listener garbage, or discovering active teammates and their status. Triggers: 'rhizo', 'coordinate with the other terminal/agent', 'talk to agent', 'send task to', 'ask the other assistant', 'inter-agent chat', 'lock file', 'lock resource', 'mutex lock', 'prevent concurrent edits', 'fencing token', 'monotonic counter', 'zombie writes', 'work queue', 'enqueue task', 'claim task', 'ack task', 'reliable queue', 'dead letter queue', 'dlq', 'blackboard', 'scratchpad', 'shared memory', 'floor control', 'speaker ring', 'moderated roundtable', 'pass floor', 'yield floor', 'cancel task', 'cancel run', 'cancellation token', 'abort run', 'ballot', 'vote', 'consensus', 'blind voting', 'leader election', 'acquire leader', 'failover', 'mesh leader', 'scatter', 'gather', 'quorum', 'workflow', 'dag', 'pipeline', 'task dependencies', 'resolve step', 'workflow next', 'sweep', 'clean dead agents', 'stale locks', 'garbage collection', 'cluster watchdog', 'who is online', 'agent status', 'rhizo', 'Redis bus'.
Use when accepted ADRs exist and AD.md must be produced or updated as unified architecture documentation.