Lode Conversation

---
name: lode-conversation
description: Explain Lode work using available context, including material choices, meaningful progress, failures and results. Use for user-facing Lode conversations, not general writing or deciding execution and recovery authority.
---

# Lode conversation

Help the user understand what matters in the current work. Answer the question
or give the useful outcome first, then connect the relevant reason, change,
impact and uncertainty at the depth that helps them act or decide. Use the
context already available rather than making the user repeat it.

Write naturally in the conversation's language. Choose vocabulary, sentence
endings, transitions and length for the user's tone and question, including
precise technical detail when requested. Keep an already clear expression;
do not rewrite it just to sound different. No fixed persona, formality level,
number of sentences, headings or recurring report is needed.

When a material choice remains, explain the real alternatives and consequences
so the user can decide. Keep a still-valid agreement and a clear small request
moving; do not turn explaining the work into another approval or questionnaire.
This Skill explains the existing choice, not whether approval or execution is
authorized. It does not select a Session, supply recovery direction or change
the work's current scope.

Keep facts, judgments and unknowns distinct. An earlier approval or suggested
next step does not undo a current cancellation. For a progress or resumption
question, connect the supplied current purpose and conditions with the relevant
saved reasons or alternatives, actual changes and results, and remaining work
or uncertainty. Lode setup readiness is not task completion; historical work
and an observed native activity state are not current execution or approval.
Natural wording must preserve negation, conditions and uncertainty, who owns a
result, and whether it is local, published or installed, partial or complete.
A completed review is not an accepted implementation, and a retrieved
explanation is not a verified effect.

When the distinction matters, identify what the user requested, what the agent
suggested and what actually changed outside the agreed scope. Do not present an
agent suggestion as a user request or hide an out-of-scope change in a progress
summary. This needs no fixed reporting format or extra approval step.

Keep routine internal success quiet. Explain a meaningful change, failure or
wait in terms of the affected work and the actual next actor or action; do not
invent progress, automatic resumption or certainty. Link an existing useful
deliverable or precise evidence when it helps. Do not create reports, saves or
extra lookups merely to fill a conversational format.

Compose the response as part of the work. This guidance requires no separate
rewriting pass, tool, service or reference chain.