Skip to content

Learner evidence template

Copy this template into the private Hacking/Deadwire/learning/ area for each bounded session. Keep the file private when it contains lab identifiers or source-specific details. The template separates what a source says, what the lab observed, and what the learner can explain.

# Session: YYYY-MM-DD / concept-id

## Scope and authorization

- Authorization owner:
- Lab boundary:
- Abort condition:
- Cleanup or rollback:

## Source and prediction

- Source reference(s):
- Concept ID:
- Prediction before reading or testing:

## Environment

- Host or snapshot identifier (redacted):
- OS, tool, and version:
- Account or role (redacted):
- Network boundary:

## Bounded action

```text
Paste only the minimum command or procedure needed for this lab step.
Mark it unverified until the evidence below supports it.

Observation

  • Timestamp and timezone:
  • Redacted output or error:
  • What this directly proves:
  • What it does not prove:

Interpretation and defense

  • Mechanism in my own words:
  • Detection or defensive control:
  • Blind spot or alternative explanation:

Scores

  • Recall (0–1):
  • Practical execution (0–1):
  • Explanation (0–1):
  • Evidence reference for lab/explanation scores:

Reflection

  • What failed or surprised me:
  • Next bounded experiment or review:
  • Reviewer and review date: 
    ## Promotion rule

Record separate learning events only after the evidence is saved. A lab or explanation event requires a redacted evidence reference:

```bash
deadwire record-learning event-YYYYMMDD-01 concept-id lab 0.8 \
  --evidence-ref learning/session-YYYY-MM-DD-concept-id.md \
  --state-path learning/state.json

Do not record a score for an action that was not run. A failed experiment can still receive an explanation score when the learner accurately identifies the failed prerequisite and the next test.

Redaction checklist

  • Remove passwords, tokens, private keys, cookies, and session material.
  • Replace real hostnames, usernames, addresses, and personal data with stable labels.
  • Keep source paths and hashes only in private notes when they identify licensed material.
  • Confirm the lab was owned or explicitly authorized before sharing any transcript.