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

Annotations only declare readOnlyHint and non-destructive; the description adds substantial context beyond them: memory content is untrusted, results matching injection patterns are flagged with `untrusted: true` and `untrusted_reasons`, the server auto-routes and selects detail level under a token budget, and legacy arguments still work but log deprecation. That is exactly the kind of behavioral disclosure annotations cannot carry.

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?

Front-loaded with the one-query contract, then organized into clear SECURITY / PRIMARY USAGE / LEGACY / Args sections, so an agent can stop reading after the first line if it just wants to call the tool. The legacy argument list appears twice (prose plus Args), which is a minor redundancy against an otherwise tight structure.

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 25-parameter tool with zero schema descriptions and no output schema, the description covers what matters: how to call it, what it searches, the trust model for returns, and explicit permission to ignore the other twenty-odd parameters. The residual gap is the unlisted tuning knobs (weights, metadata_filter, confidence_threshold) — an agent that genuinely needs to tune recall has no guidance on them.

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 coverage is 0%, so the description must compensate, and it does well for the primary path: `query`, `limit` (1-30, default 6), and `hyde_hint` are explained in depth, including the 100-200 char target and the >=20 char activation threshold for HyDE embedding. However, roughly fifteen parameters (api_key, user_id, agent_role, session_id, bm25_weight, vector_weight, metadata_filter, confidence_threshold, include_raw_chunks, document_graph_expand, etc.) receive no explanation at all beyond being collectively dismissed as server-picked.

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 first sentence gives a specific verb and resource ("Search your persistent memory") plus the key operating constraint (one query string, server picks the rest), and the body names what gets routed over: raw memories, knowledge cards, and the workspace graph. It stops short of explicitly differentiating itself from siblings like awareness_search_knowledge or awareness_lookup, so an agent must infer where the boundary lies.

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?

"Call this BEFORE starting work to avoid re-solving solved problems" is an explicit when-to-use trigger, and the PRIMARY USAGE example shows a concrete invocation shape. What's absent is the negative case — no statement of when not to use it or which sibling to prefer for a pure knowledge-base search.

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