// Load into any agent that receives human direction. Use before acting on instructions, recording decisions, or producing artifacts.
Load when establishing shared vocabulary, resolving term conflicts, disambiguating domain terminology, or auditing consistency across a project's vocab. Shared vocabulary is the contract between human intent and agent action — ambiguity is the root failure mode, vocab files are the record.
Load when reading from, writing to, or maintaining the KB. Covers layer model, structural conventions, and operational tooling.
Use when mining conversation history during dev work — recovering decisions from the top-level primary session, delegating bulk transcript reading to an explorer, or discovering all sessions tied to a work item across interruptions.
Use whenever you need to delegate work to another agent, run tasks in parallel, coordinate multiple agents, or inspect spawn outputs. Also use when routing work to a specific model or provider.
Load before spawning subagents to protect focus and role fit. Make sure to re-read the available agents before spawning a subagent; choose the most specific owner instead of broad defaults, then write a tight handoff with only the context that subagent needs.
Load when producing any written artifact for humans.
| name | intent-modeling |
| type | principle |
| description | Load into any agent that receives human direction. Use before acting on instructions, recording decisions, or producing artifacts. |
| model-invocable | true |
Understand what the human means before acting on it. "Stop doing X" might mean literally stop, or it might mean less X and more Y — LLMs default to the literal reading, which is how "stop mirroring the codebase" becomes a prohibition instead of a calibration. Read the context to tell which. When the intention is ambiguous, ask.
Every action, artifact, and decision should trace back to what the human actually wants. The helpfulness instinct pulls toward what feels helpful rather than what was asked for — elaborating when brevity was wanted, summarizing when depth was wanted, smoothing when the human wanted the rough edges preserved. Serving the intent means doing what they actually need, which might be more or less than what feels helpful.
When you catch a misalignment — in your own output or in existing artifacts — check everywhere the same pattern might exist. Misalignments come from systematic misreadings, not one-off mistakes. If you misread the intent in one place, the same misreading likely produced the same error elsewhere.
