agentleFS
solo-istsolo-ist/prose/.claude/skills/pipeline-pe/SKILL.md

pipeline-pe

Skill8 starsChanged 5 months agoMITView on GitHub

Principal Engineer architectural risk analysis for PRs. Assesses blast radius, hidden coupling, and privilege boundary concerns. Outputs LOW-CRITICAL risk rating.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    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:

    Ask yourself: "If this change has a subtle bug, what's the worst that happens?"

    line 59
  2. Distinctive:

    When in doubt, lean MEDIUM — the cost of over-escalating is a human glance.

    line 64
  3. Distinctive:

    Check changed file paths against privilege boundary list

    line 70

≈ “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

Pipeline PE (Principal Engineer) Analysis

Architectural risk assessment from a senior engineering perspective.

Usage

/pipeline-pe <pr-number>

Analysis Framework

The PE agent evaluates changes across five dimensions:

DimensionWhat It Catches
Architectural coherenceDuct tape vs proper integration
Blast radiusFailure propagation paths
Root cause vs symptomBand-aids vs real fixes
Tech debt trajectoryIs the codebase getting better or worse?
Hidden couplingImplicit contracts created or broken

Electron-Specific Checks

CheckWhy It MattersExample
Process boundary respectRenderer <-> main must go through IPCImporting fs in renderer bypasses sandbox
Context bridge exposureNew preload APIs need input validationAdding window.api.exec() without sanitization
getApi() abstractionDirect window.api bypasses the browser fallbackShould always use getApi() from browserApi.ts
IPC validationHandlers must validate renderer inputvalidatePath() on all filesystem operations
Store interactionsCross-store deps create hidden couplingchatStore reading from editorStore directly

Risk Levels

LevelMeaningRouting ImpactExample
LOWStandard changes, well-containedhitl-light (human or Oz agent reviews)Fixing a typo in a UI string
MEDIUMSome architectural concernhitl-full (if score ≥ 4) or hitl-light (if score ≤ 3)Adding a new Zustand store with cross-store subscription
HIGHSignificant risk, multiple concernshitl-fullChanging the IPC channel protocol
CRITICALSecurity, data loss, or systemic riskhitl-full (always)Weakening CSP, disabling contextIsolation

Privilege Boundary

PRs touching these paths are auto-flagged as privileged: true:

  • src/main/** — full system access (Node.js, filesystem, network)
  • src/preload/** — context bridge / attack surface
  • electron-builder.* — build/packaging config (signing, fuses, entitlements)
  • electron.vite.config.* — build config (define blocks, entry points)

privileged: true is a routing directive, not a bug. It forces hitl-full regardless of score. This is correct — changes to the trust boundary between renderer and main process should always have human review.

LOW vs MEDIUM Decision Guide

Ask yourself: "If this change has a subtle bug, what's the worst that happens?"

  • LOW: User sees a visual glitch, a tooltip is wrong, a log message is malformed. The app still works. Fix is straightforward and contained.
  • MEDIUM: User data could be stale, a sync could skip items, a store subscription fires out of order. The app works but produces wrong results in some cases. Fix requires understanding interaction between components.

When in doubt, lean MEDIUM — the cost of over-escalating is a human glance; the cost of under-escalating is a shipped bug.

Workflow

  1. Fetch the review comment and changed files from the PR
  2. Send to Claude Opus with the PE analysis prompt
  3. Check changed file paths against privilege boundary list
  4. Produce risk assessment, concern list, and verdict
  5. Output structured analysis with sentinel for downstream pipeline

Output Format

Machine-readable sentinel: <!-- pe-output: {"risk":"LOW|MEDIUM|HIGH|CRITICAL", "privileged": true|false, "concerns": N} -->

Human-readable analysis with architecture review, specific concerns, and verdict.

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.