skills/checkpoint/SKILL.md
Snapshot the current belief state (and optionally git-stash the working tree) under a label, so a risky action can be rolled back if it produces high surprise or breaks acceptance tests.
npx skillsauth add jason-hchsieh/predictive-mind checkpointInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Creates a two-layer checkpoint you can roll back to:
belief-store.snapshot_beliefs — preserves
the inspectable belief state at this moment under a label.git stash push -m <label> — preserves
uncommitted changes so a risky action can be undone.Use before:
risk_flags.<label> — 1-3 words, no spaces.belief-store.snapshot_beliefs(task_id, label). Capture the
returned snapshot_id.git status --short), run:
git stash push -u -m "pm-checkpoint:<label>".belief-store.rollback_beliefs(task_id, label).git stash list | grep "pm-checkpoint:<label>".git stash apply stash@{N} (prefer apply over pop so you
can re-checkpoint the rollback itself)./probe (e.g. git status) to confirm the rollback landed as
expected.git stash drop stash@{N}. Don't let checkpoint stashes
accumulate.$ARGUMENTS — the label (required). If absent, use an ISO-timestamp
derived label (e.g. ckpt-20260101-1200).
tools
Surprise-triggered reflection. When a PostToolUse hook reports high prediction error (surprise >= 0.6), invoke /reflect to update beliefs via the belief-reviser and decide whether to continue, replan, reclarify, or ask the user.
testing
Run a cheap, reversible real probe to ground a belief instead of hallucinating. Use when dreamer confidence is low or when you catch yourself guessing about repository state.
tools
Record an explicit predicted observation before a side-effecting tool call. Use whenever you are about to run a command, edit a file, or invoke any tool whose outcome you have not already observed. Upholds the predict-before-act invariant.
development
Given a goal and a set of candidate next actions, rank them by Expected Free Energy (pragmatic + epistemic value) using past similar episodes, and pick the winner. Delegates scoring to the policy-selector subagent.