richfrem/agent-plugins-skills / plugins/agent-orchestration/.apm/skills/orchestrating-agent-work
orchestrating-agent-work
VerifiedInteractive orchestrator for complex engineering work. Analyzes the task, conducts a diagnostic interview to select the appropriate execution strategy (Directed Acyclic Graph versus Loop, and which Loop type), plans the artifact, and supervises execution and verification. Use for "how should I orchestrate this", "plan this workflow", "should this be a DAG or a loop", "run these jobs in parallel", "delegate this to a worker agent", or "run/resume this graph manifest". Not for single trivial edits or skill self-improvement.
references/graph-manifest-schema.json
View on GitHub{
"$schema": "http://json-schema.org/draft-07/schema#",
"title": "GraphExecutionManifest",
"description": "Schema for token-budgeted, deterministic execution graphs (DAGs) compiled by graph-planner and consumed by graph-execution.",
"type": "object",
"additionalProperties": false,
"required": ["schema_version", "graph_id", "version", "budget", "nodes"],
"properties": {
"schema_version": {
"type": "string",
"pattern": "^\\d+\\.\\d+\\.\\d+$",
"description": "Semantic version of the manifest schema contract."
},
"graph_id": {
"type": "string",
"pattern": "^[a-z0-9]+(-[a-z0-9]+)*$",
"description": "Kebab-case task or workflow identifier."
},
"version": {
"type": "string",
"pattern": "^\\d+\\.\\d+\\.\\d+$",
"description": "Semantic version of the graph definition."
},
"caller": {
"description": "Opaque metadata or trace information supplied by caller environment.",
"oneOf": [
{ "type": "string" },
{ "type": "object" }
]
},
"budget": {
"type": "object",
"additionalProperties": false,
"required": ["token_ceiling_total", "max_parallel_concurrency", "max_retries_per_node"],
"properties": {
"token_ceiling_total": {
"type": "integer",
"minimum": 1000,
"description": "Advisory token budget ceiling allocated across all graph nodes."
},
"max_parallel_concurrency": {
"type": "integer",
"minimum": 1,
"maximum": 16,
"description": "Maximum concurrent parallel workers allowed during fan-out."
},
"max_retries_per_node": {
"type": "integer",
"minimum": 0,
"maximum": 2,
"description": "Maximum retry attempts before triggering failure escalation (max 2 retries = 3 attempts total)."
}
}
},
"nodes": {
"type": "array",
"minItems": 1,
"items": {
"type": "object",
"additionalProperties": false,
"required": ["id", "type", "tier", "command_or_eval", "input_bindings", "output_contract", "depends_on", "failure_escalation"],
"properties": {
"id": {
"type": "string",
"pattern": "^[a-zA-Z0-9_-]+$",
"description": "Unique identifier for this node."
},
"role": {
"type": "string",
"enum": ["approval", "worker", "verifier", "barrier", "fallback"],
"description": "Functional role of node. 'approval' identifies leading approval verifier_gate."
},
"type": {
"type": "string",
"enum": ["parallel_read", "sync_barrier", "verifier_gate", "sequential_mutation"],
"description": "Execution semantics and boundary behavior."
},
"tier": {
"type": "string",
"enum": ["fast_engine", "frontier_engine", "deterministic_script"],
"description": "Compute tier assigned based on token economics."
},
"command_or_eval": {
"type": "string",
"description": "Shell command executed with bash -c in the workdir. Model-tier nodes invoke their model through a CLI inside this command."
},
"timeout_seconds": {
"type": "integer",
"minimum": 1,
"default": 300,
"description": "Maximum execution time in seconds before terminating node process."
},
"verifier_hash": {
"type": "string",
"pattern": "^[a-f0-9]{64}$",
"description": "Expected SHA256 hex digest of the first file named in command_or_eval, or of the command string when it names no file."
},
"mutation_targets": {
"type": "array",
"items": { "type": "string" },
"description": "Explicit relative file paths this mutation node is permitted to modify."
},
"forbidden_paths": {
"type": "array",
"items": { "type": "string" },
"description": "Glob patterns that this mutation node must never modify."
},
"reversible": {
"type": "boolean",
"default": true,
"description": "Whether mutations or actions by this node can be undone via git rollback. External network/API calls must set false."
},
"input_bindings": {
"type": "array",
"items": { "type": "string" },
"description": "Upstream node outputs bound as inputs. Declarative: the runner does not pass data; nodes read sibling scratch directories."
},
"output_contract": {
"oneOf": [
{
"type": "string",
"description": "Path to schema JSON or format descriptor defining node output."
},
{
"type": "object",
"description": "Inline JSON schema object defining node output contract."
}
],
"description": "Output validation contract as an on-disk schema path or an inline JSON schema object. Declarative: the runner judges nodes by exit code; enforce contracts in a verifier_gate or sync_barrier command."
},
"depends_on": {
"type": "array",
"items": { "type": "string" },
"description": "IDs of prerequisite nodes that must complete before this node executes."
},
"failure_escalation": {
"type": "string",
"enum": ["rollback_and_halt", "fallback_node"],
"description": "Action taken when node execution or verification exhausts all retries."
},
"fallback_node_id": {
"type": "string",
"description": "Target fallback node ID required when failure_escalation is fallback_node. The target must have role 'fallback' and the same type as this node."
}
},
"allOf": [
{
"if": {
"required": ["failure_escalation"],
"properties": { "failure_escalation": { "const": "fallback_node" } }
},
"then": {
"required": ["fallback_node_id"]
}
},
{
"if": {
"required": ["type"],
"properties": { "type": { "const": "verifier_gate" } }
},
"then": {
"properties": { "tier": { "const": "deterministic_script" } }
}
},
{
"if": {
"required": ["role"],
"properties": { "role": { "const": "approval" } }
},
"then": {
"properties": { "type": { "const": "verifier_gate" } }
}
},
{
"if": {
"required": ["type"],
"properties": { "type": { "const": "sequential_mutation" } }
},
"then": {
"required": ["mutation_targets"]
}
}
]
}
}
}
}
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |