agentleFS
jiayuewangjavyjiayuewangjavy/product-designer-skills-os/skills/intent-preservation-check/SKILL.md

intent-preservation-check

Skill4 starsChanged 20 days agoMITView on GitHub

Compare an AI-generated product revision with confirmed product intent, constraints, and prior decisions to detect drift. Use after substantial changes to a brief, flow, prototype, interface, or implementation. Do not use to block intentional changes that have explicit evidence and ownership.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 5 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Keep assumptions and preferences separate from confirmed decisions.

    line 14
  2. Distinctive:

    Do not invent baseline intent when no trustworthy source exists.

    line 43
  3. Distinctive:

    Do not label every difference as drift.

    line 44
  4. Distinctive:

    Do not silently revert, overwrite, publish, or deploy an artifact.

    line 45
  5. Distinctive:

    When sources conflict, expose the conflict and identify the decision owner.

    line 46

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Intent Preservation Check

Detect whether a new artifact silently changed decisions that humans had already made.

Establish the comparison

  1. Identify the baseline artifact or decision record and the revised artifact.
  2. Extract confirmed intent: target user and job, product promise, conceptual model, terminology, critical task, constraints, quality bar, permissions, edge states, non-goals, and explicit human decisions.
  3. Keep assumptions and preferences separate from confirmed decisions; they may change without constituting a violation.
  4. Compare observable behavior and meaning, not only words or visual similarity.

Classify changes

For each meaningful difference, classify it as:

  • Preserved: intent remains intact despite implementation differences.
  • Intentional change: evidence, owner, and rationale are recorded.
  • Unresolved: the change may be valid but lacks a decision.
  • Drift: confirmed intent was removed, contradicted, or obscured without authorization.

Prioritize drift affecting user control, permissions, safety, trust, recovery, product meaning, or the primary task. Do not demand pixel-level sameness unless visual precision was an explicit constraint.

Output

## Baseline and revision
## Confirmed intent inspected
## Change table
| Element | Baseline | Revision | Classification | Impact | Evidence |
## Critical drift
## Human decisions required
## Safe corrections
## Context or decision-record updates

Boundaries

  • Do not invent baseline intent when no trustworthy source exists.
  • Do not label every difference as drift; preserve room for implementation judgment and evidence-backed evolution.
  • Do not silently revert, overwrite, publish, or deploy an artifact.
  • When sources conflict, expose the conflict and identify the decision owner.
  • A passing check means known intent was preserved, not that the product is usable or valuable.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.