Harness task — the full mission contract - PlungeAI

Full field reference

Field	Type	What it bounds
`goal`	string	The per-run input — becomes the agent's prompt
`mission`	multi-line string	Standing purpose → system-prompt frame. A SINGLE-TOKEN string is a card reference instead (below)
`persona`	string	One voice per run (`digital_twin` alias; `personas: [x]` → first entry)
`skills`	string[]	Instruction packs — first 5 eager, rest on demand (`skills.md`)
`experts`	string[]	Domain lenses — first 3 eager (`experts-personas.md`)
`backgrounds`	string[]	Ambient context cards — first 3 eager, injected first
`plugins`	string[]	Bundles: skill index + MCP servers + scripts (`plugins.md`)
`mcp`	string[]	MCP server ids connected in-loop
`model`, `provider`	string	Model/provider override (`models.md`)
`effort`	`quick`\|`standard`\|`deep`	Turn-budget preset: 4 / 12 / 20 turns
`max_turns`	number	Explicit loop cap (alias `max_iterations`); wins over `effort`; runtime default 50
`max_parallel`	number	Widest fan-out one `delegate` call may spawn — cap-and-refuse, never silently batched
`allowed_tools`	string[]	The tool fence (below)
`allowed_agents`	string[] \| `'all'`	Agent fence for `call_agent`
`denied_agents`	string[]	Subtracted from the agent fence
`permissions`	map tool → `allow`\|`ask`\|`deny`	Per-tool human gate: `ask` pauses the run with `needs_approval` before the tool executes; `deny` rejects the call
`success_criteria`	string[]	Self-checked statements the agent verifies before finishing
`instructions`	string	Extra author instructions, appended after skills — always lands
`python_executor`	`auto`\|`pyodide`\|`anthropic`\|`gemini`	`run_python` tier routing
`memory_owner`	string	Memory namespace override — per-bot isolation (`memory.md`)
`local`, `local_agents`	bool, string[]	Local bot: actions route to the user's own machine via the desktop daemon (cloud brain, local hands); requires a connected local node

In a saved agentic agent: memory_owner is forbidden (tenancy comes from the executing user); local / local_agents are bot-only and require agent: harness-agent.

Validation Studio applies at save (validate.ts:221-311)

goal required (prompt: is an alias); mission (text or object) OR mission_ref required.
allowed_tools must be an array; max_turns a positive number; effort ∈ quick|standard|deep.
skills / experts / backgrounds / plugins / mcp must be string arrays; persona one string.
Ids are NOT checked at save — an unknown id degrades to a warning at run time. Declare only listed ids.
Merge order when a card is referenced: card → workflow → task, last wins; arrays replace.