agentleFS
erikunhaerikunha/vault-of-loci/.agents/skills/retention-audit/SKILL.md

retention-audit

Skill0 starsChanged 5 days agoMITView on GitHub

Audit the learning ledger for objectives that passed the 7, 30 and 90 day horizons with no attempt without help, write a new closed book task for each one that is due, and record the confirmed attempt. Use when the learner asks what is due for a delayed test. Writes a file, so only run when asked.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Run the audit and read it.

    line 17
  2. Very specific:

    Use the item id the audit prints for that objective, and the mode of the objective from 90_System/Objectives.json.

    line 38

≈ “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

Tutor mode: the learner attempts first, closed book. The agent never writes the learner's notes, answers or guesses. Hints come one step at a time and only after an attempt. Confidence from 1 to 5 is asked before any reveal. The agent never grades an attempt the learner did not confirm.

This skill answers "what is due for a delayed test". It does not answer "what should I study today", which is schedule_review.py. The two sound alike and give different answers: this one names the objective that passed a 7, 30 or 90 day horizon with no attempt without help, and the other ranks what is worth another try. If the learner asked what to study today, use the other skill.

  1. Run the audit and read it. It reports the objectives that passed a horizon with no attempt without help, and the objectives that were never tested.
python3 -I 90_System/Scripts/retention_audit.py
  1. Tell the learner which objectives are due. Say in one line that the delay has arrived but the retention at that horizon is still unknown until the attempt happens, and that an attempt with a hint stays out of the rate. To see only what is due, to change the horizons, or to see the audit as data:
python3 -I 90_System/Scripts/retention_audit.py --plan
python3 -I 90_System/Scripts/retention_audit.py --horizons 7,30,90
python3 -I 90_System/Scripts/retention_audit.py --json
  1. For each objective that is due, write a new task the learner has not seen. Build it from the objective note and the sources it cites in the vault. Never reuse the example that was studied, and never take the task from memory of the conversation.
  2. Use the item id the audit prints for that objective, and the mode of the objective from 90_System/Objectives.json.
  3. Give the task to the learner and wait. The attempt is closed book: no notes, no excerpt, no summary before the answer. If the learner asks for help after the attempt, use the hint ladder one step at a time: hint, strong hint, partial example, answer. Never skip a step.
  4. Grade the learner's words against the source and the rubric. Ask for confidence from 1 to 5 before the reveal.
  5. Record the attempt only after the learner confirms the result, with the item id the audit printed. If the learner took a hint, pass --assistance and --hints: an attempt with help stays out of the rate.
python3 -I 90_System/Scripts/record_attempt.py --objective <id> --item <id> --mode <mode> \
  --correct --confidence <1-5> [--hints <n>] [--assistance <level>] [--grading <method>] \
  [--rubric <ver>] [--revision <rev>] [--minutes <n>] [--at <iso>] [--probability <0-1>]
  1. Show the learner the line that was written. Never edit or delete a line of the ledger.

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.