agentleFS
loremaster-ailoremaster-ai/loremaster/skills/impact-analysis/SKILL.md

impact-analysis

Skill4 starsChanged 4 months agoMITView on GitHub

Computes the ripple radius (affected documents and concepts) of a proposed planning change via the graphify graph BEFORE the change is made and surfaces it for human review — triggers on "what's affected if we change X?" or "let's change X to Y" (analysis/finding by default and the human judges the direction; if the user explicitly approves applying the change, apply it and then always commit and push — never leave it uncommitted); this channel = this project only.

  • Commits and pushes

11 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

2 rules. 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 10 edited versions we read keep these.

  • 2

    Distinctive

    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:

    Flag ripple points (conflict): check which points would become contradictions after the change, from the conflict-detection perspective — …

    line 31
  2. Distinctive:

    Never assert impact without evidence (a graph edge / a source document).

    line 37

“Common” is measured: kept by at least 6 in 10 of the 10 edited versions of this file we could read. “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

Change Ripple (Impact Radius) Analysis

Big planning changes are not local (a billing change → GTM, onboarding, a cascade of related features). Your job is to find and surface what will be affected. Fixing and judging are human work.

Behavior

  1. Compute the impact radius (graphify): query the graph for the change topic/node —
    • graphify explain "<topic>" : that node's neighbors (directly connected concepts/documents) + their relationships.
    • graphify path "<A>" "<B>" : the connection path between two points.
    • Identify wide-area impact via the god nodes (heavily connected hubs) and cross-links in GRAPH_REPORT.md. → Collect the list of affected documents/concepts and why each is connected.
  2. Flag ripple points (conflict): check which points would become contradictions after the change, from the conflict-detection perspective — do the affected docs clash with the change topic? Quote open conflicts as they are.
  3. Organize and present: lay out the affected items and (if the change goes through) the points that would become contradictions, at a glance, so the human can decide whether to proceed with the change. (No meeting feedback loop and no meeting-agenda chaining — this skill is pre-change impact discovery only.)

Rules (invariants)

  • The human judges which way to change. Never pick the direction yourself. But once the human approves applying it, apply that decision + commit and push (the ⚠️ section below). If only analysis was requested, git stays untouched.
  • The graphify radius is not 100% accurate (leaks and false alarms are possible) → present results as "impact candidates" and say so when you are not sure.
  • Never assert impact without evidence (a graph edge / a source document).
  • This channel = this project only. This project's graph and lore vault only.

⚠️ If you apply a change — commit and push are mandatory (otherwise it is unreflected and the Obsidian pull is blocked)

This skill's default is analysis (finding). But when the user explicitly requests application — e.g. "let's change Google → Naver / change it for me" — and presses the approval button (= the human has already decided), apply the change and then always land it in git before finishing. Applying without committing leaves the change floating in the working tree only — it never reaches origin or the wiki — and the pull from a separate clone (Obsidian) is blocked by the dirty tree.

  1. Patch raw/PRD.md (+ directly affected docs). Prefer not to edit the wiki directly (refine derives it from raw) — patch raw only. If you had no choice but to edit the wiki directly, commit that too.
  2. If the user explicitly named specific Jira backlog updates along with the PRD change, update those Jira issues via the REST API within the same approval scope and verify by re-fetching them. If a Jira issue key was mentioned, prefer a Jira lookup over document search. Jira edits are an external write separate from the docs commit — confirm they are inside the approval scope, and if not, get a separate request_approval.
  3. Immediately after applying, always run git add -A <vault> && git commit (even a local commit alone leaves the working tree clean, so the Obsidian pull is not blocked).
  4. The origin push goes through a request_approval button (title example: origin push — <project> (change applied)). APPROVED → git push origin main. DENIED / no response → hold only the push (keep the local commit).
  5. If the PRD change makes the current spec in existing wiki/concepts/* stale, load the refine-wiki skill after landing raw and refresh the affected concept pages, wiki/index.md, and the graphify outputs. This refinement is derived from the raw decision, so it needs no fresh content approval — but after its separate commit, the origin push again goes through a request_approval button. In particular, for concepts graphify links directly to the changed topic (e.g. a change touching pricing, data model, navigation, or a related recommendation feature), check the wiki for staleness as well. Never end with an applied change left uncommitted. (Analysis-only runs leave git untouched — nothing was changed.)

Output

  • Change topic → impact radius (documents/concepts + why each is connected) + points that would become contradictions if changed. Slack mrkdwn sections, not tables.

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.