recursive_agent_execute

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations, the description carries the full burden and does substantial work: it discloses authority inheritance (tenant/company/project/workspace), disjoint rollout allocation, root vs. child budget fraction semantics, per-child compaction plus tree-wide context-byte caps, the allowlist mechanism, and Spring's re-check of runtime workspace scope and approval. This is rich governance context beyond anything the schema exposes, though it stops short of describing what happens on budget exhaustion or whether execution is asynchronous.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is front-loaded with the action sentence and every subsequent sentence earns its place by explaining governance semantics. It is dense (~130 words) but not padded. It loses a point for reliance on unexplained jargon ('disjoint finite rollout allocation', 'RLM context chunking', 'provider budget') that an agent may not decode on first read.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a high-complexity tool with 15 parameters and zero annotations, the description covers the hardest dimensions well: authority inheritance, budget reservation, delegation context caps, allowlist restriction, and workspace approval revalidation. The output schema covers return values. Gaps remain around the execution lifecycle (how execution_id is obtained, status/cancel companion tools) and what 'REPL work' concretely entails, but the core correctness-critical semantics are present.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 0%, so the description must compensate. It adds real meaning for some of the most opaque parameters: root_budget_fraction (root's initial share of the tree-wide provider budget), allowed_agent_ids (comma-separated allowlist semantics), and the two context-byte settings (per-child compaction and tree-wide cap). But most of the 15 parameters (execution_id, inputs format, max_total_nodes, max_children, max_repl_steps) receive no semantic explanation beyond their names.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Does the description clearly state what the tool does and how it differs from similar tools?

The opening sentence states a specific verb and resource: 'Run a true recursive Lightbulb agent with bounded subagents and REPL work.' It also distinguishes the approach from 'legacy RLM context chunking,' which helps an agent understand this is genuinely recursive execution rather than a chunking wrapper. It falls short of 5 because it never names the sibling execution tools (backbone_execute, dispatch_domain_agent) it should be compared against.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description gives some contextual signals, notably 'This is distinct from legacy RLM context chunking' and a conditional hint for allowed_agent_ids ('when only specific marketplace/domain workers may be delegated'). However, there is no explicit when-to-use versus when-not-to-use guidance naming sibling tools, and no mention of the execution lifecycle partners (new_recursive_execution_id before, get_recursive_agent_execution_status after).

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.