agentsclimarketplace

Nd selfmonitor

Skill jpoindexter/nd-skills/skills/nd-selfmonitor

Before destructive or hard-to-reverse operations, pause to verify the action matches the stated goal. A one-sentence pre-flight check, not a gate.From its SKILL.md

Install
npx -y skills add jpoindexter/nd-skills --skill nd-selfmonitor

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • 27 days oldThe repository was created 27 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

2.8 KB, 577 tokens by cl100k_base, as published. Nobody here has run it

Self-Monitor — Pre-Action Intent Check

What this addresses

Self-monitoring: the executive function that asks "is what I'm about to do actually what I meant to do?" This is a brief meta-cognitive step that normally happens automatically before consequential actions. In ADHD and high-focus states, it can be skipped — leading to "I don't know how I got here" moments where a command was run or a file was overwritten unexpectedly.

The self-monitor skill inserts a visible intent-check before high-consequence actions.

When to apply

Apply before:

  • Running destructive shell commands (rm, drop, truncate, git reset --hard, force push)
  • Database mutations (DELETE, DROP, UPDATE without WHERE)
  • Committing code (verify the right files are staged, the message matches the work)
  • Closing or deleting items from a tracked backlog
  • Replacing a user-owned file wholesale when the change is hard to review or restore

Do the check silently for routine, reversible in-scope edits. A patch to a tracked file is not automatically a user-facing interruption.

Instructions

Before any of the above operations, emit a one-sentence pre-flight check:

→ About to [action] — this serves [current goal] because [reason]. Proceeding.

Examples:

  • "→ About to overwrite auth.ts — this serves 'fix the login bug' because the session token logic needs replacing. Proceeding."
  • "→ About to run git reset --hard HEAD~2 — this serves 'undo the broken migration' because both commits are in the bad batch. Proceeding."

If the reason feels unclear when writing it: Do not proceed. Surface the uncertainty:

"I was about to [action] but I'm not certain it serves [goal]. Want me to check before I run it?"

The check is internal by default. Show it only when:

  • The action is irreversible (cannot be undone without effort)
  • The user would reasonably want to know
  • Something feels misaligned
  • A commit or external state change is about to bundle multiple actions; show one check for the batch, not one per file

Science basis

Self-monitoring is the meta-cognitive component of executive function: evaluating one's own actions against intention in real time. It's distinct from planning (deciding what to do) and from task execution (doing it) — it's the bridge between them.

In ADHD research, impaired self-monitoring is associated with impulsive errors: correct intention, wrong execution. The intervention is an externalized check that makes the intent explicit at the moment of action — so a mismatch becomes visible before the action is taken.

The "proceeding" ending is intentional: the check should not add friction to correct actions. Only uncertain actions get a pause.

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Keep looking

Skills are one crate of 326,852. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.