friction_scan

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

Beyond the annotations (readOnlyHint=false, destructiveHint=false), the description reveals critical behaviors: it only flags (never blocks), it persists and dedupes flags, it is model-free and deterministic (no LLM, no egress), and it must be driven externally. These details significantly inform the caller about side effects and operational constraints, going well beyond the annotation hints.

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 notably verbose and repetitive. It restates the same idea in multiple ways (e.g., 'agent has stopped being *other*' and 'naming where the agent stopped disagreeing'), and includes extensive meta-commentary about model-free operation, non-blocking behavior, and external invocation. While the main purpose is front-loaded in the first sentence, the overall length and redundancy detract from clarity and maintainability.

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?

The description gives substantial context about the tool's domain and side effects (flagging, persistence, deduping), but it omits critical details: it does not define the output/return value (e.g., success flag, list of flags created), does not specify error conditions, and leaves parameter semantics incomplete (especially app_id and the precise meaning of floor/window). Without these, the tool is not fully self-contained for an agent to use reliably.

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?

The description partially explains parameters: 'turns' is explicitly defined as a list of role/text/timestamp objects, and 'floor' and 'window' are referenced as 'friction floor' and 'window of agent turns'. However, the meaning of the numeric 'floor' threshold and the exact role of 'window' in the scan are not precisely clarified, and the required 'app_id' parameter is entirely unexplained. Schema coverage is 0%, so the description adds some value but leaves gaps.

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 description clearly states the tool's purpose: scanning a transcript window for the 'mirror failure mode' where an agent reflects the user while the user escalates. It also specifies key constraints (deterministic, no LLM, non-blocking) and distinguishes itself from a listing tool like friction_flags_list by focusing on detection rather than enumeration.

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 provides implicit usage context: it is a deterministic checker that flags issues without blocking, and it must be invoked from outside the watched model (harness/monitor). However, it does not explicitly state when to prefer this tool over alternatives (e.g., when to use friction_scan vs. friction_flags_list), nor does it outline clear conditions for invocation beyond the general scenario.

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