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

With no annotations, the description carries the full burden and does well: it discloses there is no automated fuzzy matching, that the agent must check evidence itself, that unrelated content is preserved, and crucially that it 'returns instructions for your own editing tools' and 'never writes files'. It omits permissions, concurrency, or reversibility details, but the non-destructive, advisory nature is clearly established.

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?

Two dense sentences that lead with the action and the trigger before the mechanics. Every clause carries information (no fuzzy matching, returns instructions, preserves content), though the second sentence stacks several clauses and could be split for readability.

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 two-parameter mutation-adjacent tool with no output schema and no annotations, the description covers the important unknowns: what it returns ('instructions for your own editing tools'), that it is advisory rather than destructive, and that unrelated content is safe. Only minor gaps remain around cwd semantics and error behavior.

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 already 100% for both parameters, so the schema carries the semantics. The description reinforces the 'describe the facts in natural language' contract for the description parameter but adds no syntax or format guidance for cwd. Baseline 3 is appropriate.

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?

States a concrete action on a concrete resource ('Correct or remove stored facts') and the trigger condition. However, the name says 'forget' while the description broadens the scope to 'correct', and it later reveals it does not actually delete anything — creating a slight mismatch an agent must reconcile. Sibling tools memory_save/memory_review are not explicitly contrasted.

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?

It gives clear triggering conditions ('when evidence shows they are wrong, superseded, or obsolete') and an imperative ('promptly'), which tells the agent when this tool is appropriate. It does not name the alternatives (memory_save, memory_review) or state when NOT to use it, so routing between siblings is left implicit.

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

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

With no annotations present, the description carries the full burden and does so: it discloses the return payload (resolved file, word count, soft budget, evidence-based cleanup instructions) and explicitly states the tool never edits or deletes files itself. That side-effect profile is the single most important behavioral fact for a memory-management tool.

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?

Three sentences, each with a distinct job: purpose, usage boundaries, behavior/return contract. Nothing is redundant and the triggering condition is front-loaded.

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 zero-annotation, one-parameter tool with no output schema, the description covers what the tool does, when to call it, what it returns, and what it will not do. No material gap remains for correct invocation.

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?

There is one parameter (cwd) with 100% schema description coverage, so the schema already explains its meaning and the multi-root requirement. The description adds no syntax or format detail beyond that, which is the expected baseline when the schema does the heavy lifting.

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 pairs a specific verb ('review and trim') with a specific resource ('project memory') and clarifies the tool's role: it inspects and produces cleanup instructions rather than mutating. This distinguishes it from the sibling tools memory_save and memory_forget, which is exactly the routing signal an agent needs.

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?

It gives both positive triggers ('when the user requests cleanup or a major project change makes guidance obsolete') and explicit exclusions ('Do not call every turn or after routine tasks'). This is textbook when/when-not guidance and leaves nothing to inference.

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

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

With no annotations, the description carries the full burden and does well: it discloses that the tool returns a resolved path, size guidance, and merge instructions, and that it never writes files. The somewhat ambiguous final clause ('never writes files or requires saving a low-value candidate') muddies whether a write occurs, leaving a small gap.

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 usage guidance is front-loaded well, but the passage is a dense run-on paragraph whose final clause ('never writes files or requires saving a low-value candidate') is grammatically tangled and could mislead. Tightening would improve it without losing content.

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?

No output schema exists, and the description compensates by describing what is returned (resolved memory path, size guidance, merge instructions). For a 2-parameter tool with full schema coverage, little is missing; only the write-vs-resolve semantics could be stated more crisply.

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 100% for both parameters, so the baseline is 3. The description adds only marginal meaning (batching 'related' learnings maps to the learning param), but cwd and learning syntax are fully covered by the schema.

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 makes clear this is about capturing non-obvious project decisions/gotchas as memories and resolves them to a memory path, and the closing clause distinguishes its behavior from a plain file-writing save. It does not explicitly name memory_review or memory_forget, so sibling differentiation is only implicit.

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?

It gives explicit when-to-use ('at task completion only when they prevent a likely future mistake or substantial repeated work'), explicit when-not (cheaply discoverable facts, task summaries, temporary state, duplicates), and batching guidance. This is close to ideal routing guidance.

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