Correction Memory
PassAudited by ClawScan on May 1, 2026.
Overview
This skill appears purpose-aligned, but it stores corrections persistently and reuses them automatically to steer future agents.
This looks coherent for a correction-memory skill. Before installing, understand that corrections become persistent instructions for future agents, so keep them specific, do not include secrets, and review the workspace correction files if future agents start behaving unexpectedly.
Findings (2)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
A bad, stale, or too-general correction could steer future agents for up to 30 days, and any sensitive text included in a correction would be stored locally in the workspace.
The skill intentionally turns logged corrections into persistent context for future agents, which is useful but means incorrect or overly broad corrections can influence later tasks.
automatically injects it into future spawns of the same agent type
Only log deliberate corrections, avoid putting secrets in correction text, and periodically review or delete files under $OPENCLAW_WORKSPACE/memory/corrections.
Running the helper allows it to create and update correction-memory files in the chosen workspace.
The installation instructions ask the user to copy and run an included JavaScript helper. This is disclosed and central to the skill, but it is still local code execution.
cp references/correction-tracker-template.js $OPENCLAW_WORKSPACE/lib/correction-tracker.js node $OPENCLAW_WORKSPACE/lib/correction-tracker.js
Install it only in the intended workspace and inspect the helper before running it if you are concerned about local file changes.