Execute DAG workflows with concurrent agent isolation, permissions, retries, and execution tracing.
npx skills add https://github.com/curiositech/port-daddy --skill dag-runtime
Please help me install this Agent Skill. Skill: dag-runtime Source: https://github.com/curiositech/port-daddy/tree/main/skills/dag-runtime Command: npx skills add https://github.com/curiositech/port-daddy --skill dag-runtime
Orchestrates and executes planned Directed Acyclic Graph (DAG) workflows to coordinate concurrent agents, enforce isolation and permissions, and provide full execution tracing and failure recovery so teams can run complex multi-node tasks reliably.
Execute the DAG named "security-scan" with parallel waves, container isolation for risky nodes, and retry policies applied.
High-intent search queries and answers about installing and using this skill.
DAG workflow execution coordinates concurrent agent tasks by advancing nodes in parallel waves, allowing non-dependent nodes to run concurrently for faster deterministic processing. Each node spawns an isolated agent with enforced permissions and configurable retry policies.
Node isolation during parallel DAG execution supports selectable levels including container, process, context, or no isolation. Container isolation is recommended for risky nodes like untrusted code execution, balancing security with functionality while maintaining deterministic progression.
Permission inheritance in DAG agent spawning intersects child node permissions with parent node permissions to prevent permission creep. This ensures child agents never exceed the access rights of their parent nodes during parallel execution.
Failed DAG nodes support configurable retry policies with backoff and human escalation. Failures are classified by type—timeouts, schema errors, permissions, or missing skills—allowing targeted retry strategies or escalation when retries are exhausted.
DAG execution tracing records node-level metrics including model used, token counts, cost, and duration. Traced data supports execution salvage, allowing recovery of work from crashed agents for resumable workflow runs.
When agents crash during multi-wave DAG processing, execution tracing salvages recorded work from completed nodes. Partial outputs are recovered and composed into final results, enabling resumable runs without restarting the entire workflow.
curiositechPredict a five-node meta-DAG of high-impact skills from git state and project context.
curiositechInspect and replay node-level DAG execution traces to locate divergences.
ahostbrOrchestrate parallel coding agent swarms to decompose large requests into independent subtasks.
ruah-devCoordinate isolated Git worktrees and file-claim locking for parallel AI agents.
