Public-ready repository skill

Practitioner probe

Converts an ambiguous codebase concern, issue, or plausible-but-shallow implementation into a decision-changing technical probe. Applies when mechanism, affected actor, evidence, or instruction yield remains unclear.

Source and version

Repository path
.agents/skills/practitioner-probe/SKILL.md
Source revision
341bceb719a28be760775b6322021d78abaa919b
Source SHA-256
348ea113a75af3487a420e2075aa199297673aca0ba0a69e5a19d6c5a463776c
Publication standing
Public ready

Included source files

This skill has no additional files in its public source package.

Complete skill source

The Markdown below matches the exact source text. The format link above opens the file on its own.

---
name: practitioner-probe
description: Converts an ambiguous codebase concern, issue, or plausible-but-shallow implementation into a decision-changing technical probe. Applies when mechanism, affected actor, evidence, or instruction yield remains unclear.
---

# Practitioner probe

## Contract

- **What:** A small set of technical questions that distinguishes consequential possibilities.
- **When:** Code appears plausible but shallow, a gap lacks mechanism, or an issue lacks decision-changing evidence.
- **Boundaries:** A direct, already-scoped repair proceeds through its owning workflow. Scores, levels, and probabilities appear only when they change a decision. Any level or veracity label describes one artifact claim, evidence state, or repair path; it does not credential a practitioner, rank an AI system, or certify the probe itself.
- **Result:** Mechanism, observable evidence, gap utility, instruction yield, transfer boundary, and receiving route.

## Probe anatomy

1. **Question:** Consequential behavior or boundary under review.
2. **Why it matters:** Visitor, collaborator, operator, or receiving system bearing the effect.
3. **Mechanism:** Browser, cascade, schema, origin, build, data, or organizational behavior capable of producing the condition.
4. **Observable evidence:** Smallest local or remote result that distinguishes the possibilities.
5. **Instruction yield:** Next instruction that changes if the finding holds.

Source confidence and transfer limits join the probe when system, runtime, organization, or incident conditions affect applicability.

Technical-veracity labels route the next action. `correction_needed`, `local_measurement_needed`, and another claim-local state remain available when the evidence calls for them. A system's report of its own conformance remains one observation alongside source, runtime, artifact, or human evidence.

## Recovery connection

A recurring strain can be named as a recognizable system shape without assigning personal fault. Gap utility is assessed before remediation: a generative gap may remain open; a coordination or harmful gap receives the smallest adequate move routed to the surface that owns the affected state. One primary move leads, and remaining strain stays visible.

## Evolution boundary

The contract and five-field anatomy carry the current capability. Domain-specific references form only after materially different probes establish a stable mechanism family. Deterministic helpers form only when repeated mechanics benefit from exact execution. Reuse count and self-authored examples remain routing observations rather than proof of probe quality.

## Operating sequence

1. The probe pass selects the smallest decision-changing set, often one to three questions, ordered by discriminating value.
2. The owning source and one downstream consumer precede generalization.
3. A deeper probe enters only when its answer changes the receiving decision; adjacent terrain remains visibly deferred.
4. The pass stops when an instruction, source boundary, observation, loop exit, recovery move, or implementation route changes or becomes bounded.
5. An authorized direct repair routes through its owning workflow; deferred bounded work routes to an issue; durable orientation routes to a README or skill; and prospective options route to planning.

## Representative scenarios

### Scenario: Plausible CSS hides a cascade mechanism

- Given a component appears correct in source
- And one repeated visual item renders differently
- When a probe traces global rules, layout-axis behavior, and item position
- Then the mechanism identifies the smallest component-owned repair
- And the instruction yield names the affected-family observation
- And the route returns to the CSS source that owns the state

### Scenario: A repair already has clear ownership and evidence

- Given the defect, mechanism, consumer, and receiving source are already known
- When the next move is selected
- Then the owning implementation workflow proceeds directly
- And an additional probe layer remains unnecessary

## Completion

A probe completes when observable mechanism changes or confirms the next instruction. Inventory-only expansion returns for refinement or purposeful deferral.