aganesy

qfai-maintain

"Use when invoked by name or handed a QFAI work order. Use it for a change to wording, a typo, a code comment or document prose that is meant to change no behaviour."

aganesy 4 1 Updated 1d ago
GitHub

Install

npx skillscat add aganesy/qfai/qfai-maintain

Install via the SkillsCat registry.

SKILL.md

/qfai-maintain - Non-normative edits

[DRIFT-PROTOCOL:REQUIRED]

Run the entry check of
.qfai/assistant/rule/shared-skill-operating-baseline.md#workflow-run-entry-check-mandatory
first.

Inputs

  • The text or comment to change, and the write scope the request or the work
    order gives.
  • .qfai/assistant/rule/*.

Questions to the user follow
.qfai/assistant/rule/shared-skill-operating-baseline.md#user-questions-askuserquestion-protocol.

Steps

  • maintain-edit makes the edit inside the write scope, or stops before an
    edit with a semantic effect.

Run the steps as
.qfai/assistant/rule/shared-skill-operating-baseline.md#running-steps-mandatory
sets out: read .qfai/assistant/step/<name>/STEP.md for the current step only,
run it, then the next. No step is skipped.

Review

After the last step, run one review through common-review-cycle: the code
review, by implementation-reviewer. The reviewer rules on the
diff and on the no-behaviour-change judgement, and never reviews an edit it
made. A REVISE sends the finding back to the step's author.

Reviewer Gate

An edit that would change a story, a contract or a decision is not this
skill's, and stops.

Completion

Follow .qfai/assistant/rule/shared-skill-operating-baseline.md#completion-contract-shared.
Smallest applicable smoke check (this skill's override): the project's
Markdown and link checks run over the changed files, each exiting 0. A check the
project does not have is UNRUN, not a pass.

The stage returns what maintain-edit lists under "What the stage returns".
Ask for the next action only when proceeding requires the user's answer, as
.agents/rules/user-questions.md § 6 sets out. A completion-only report needs
no question. Under a no-question mode, list any remaining actions instead.

Default Autopilot Policy

  • hard-required:
    • edit target (the text or comment to change; an empty target is asked for, never guessed)

project_memory:

  • A maintenance edit changes what a reader reads, never what a program or an agent does.
  • A semantic effect found before an edit stops the edit; in a run it blocks the run, naming the owner skill.
  • The reviewer of the diff is never the agent that made it.