run_plan

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

Annotations already declare idempotentHint=true and destructiveHint=false, and the description is consistent with them while adding substantive context: revision-guarded writes via expected_revision (first 0), unique write_id dedup, size caps of 2000 UTF-8 bytes / 200 chars per item, and an explicit scope exclusion ('No strategy generation or game action'). It does not say what a conflicting revision does or what is returned, so it stops short of full transparency.

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?

Information density is high and the tool's function is front-loaded, but the text is a chain of clipped fragments and bracketed jargon ('read actually reads disk; omit other parameters', 'expected_revision(first 0)') that demands parsing. Several sentences would need rereading, which is a structure cost even if little is wasted.

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?

An output schema exists so return values need not be described, and the description covers the hard parts of a stateful read/write tool: preconditions, revision safety, dedup by ID/content, invalidation triggers, and size limits. The remaining gaps (mode value set, revision-conflict behavior) are minor given the overall coverage.

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% and no parameter has an enum, so the description carries the full burden and largely meets it: mode read/write semantics, content sub-keys (objective, priorities[1-4], recheck_when[1-4], experience_refs up to 6 read EXP-XXX@rN), observation_id freshness, expected_revision initial value, and write_id uniqueness. It still never enumerates the legal mode strings or types, leaving some inference to the agent.

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 identifies a specific resource (run memory) and distinguishes its two modes: 'read actually reads disk' versus the write path with its required fields. It is telegraphic and the opening phrase 'Model-authored short run memory' is a little opaque, but an agent can tell this is a read/write store for the current run rather than the generic read_notes/write_note siblings.

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 real preconditions ('Only confirmed start_run/continue_run, actionable current run, no pending action') and lifecycle rules ('Revise on material changes, read after compression', 'Run/profile/session changes invalidate reuse but retain history'). What is missing is an explicit contrast with adjacent tools like write_note or read_notes, so the routing to a sibling is implied rather than stated.

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