agentleFS
timurgaleevtimurgaleev/vibestack/skills/context-restore/SKILL.md

context-restore

Skill8 starsChanged 6 months agoMITView on GitHub

Restore working context saved by /context-save so you can pick up where you left off.

24 other projects use a version of this, 2 of them edited.

Adopted by

Most recently taken up by

Used with

Projects with it also use ios-clean, ios-fix and 3 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • Saqlain-811Skill

    ios-clean

    Remove the DebugBridge SPM package and all #if DEBUG wiring from an iOS app. (gstack)

    Also in 5 projects that use this

  • Saqlain-811Skill

    ios-fix

    Autonomous iOS bug fixer. (gstack)

    Also in 5 projects that use this

  • bastosSkill

    ios-design-review

    Review iOS app UI/UX design quality using Apple's design principles and the Liquid Glass-era design language from WWDC25/26. Use when asked for a design review, UX critique, visual design assessment, or evaluation of an app's interface quality, design intent, alignment with the new Apple design system, or usability. Triggers: "design review", "UX review", "visual design critique", "review the UI", "does this design work", "evaluate this interface", "design feedback", "Liquid Glass review".

    Also in 5 projects that use this

  • +2 more from Saqlain-811

The file

Show its rules and text

We split this skill into the rules it gives an agent

15 rules. 0 are common, 6 are distinctive, and 9 are very specific to this project.

  • 0

    Common

    Most of the 18 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 9

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If this session was compacted, the copy of this skill in context may be cut off.

    line 19
  2. Very specific:

    Read ${CLAUDE_SKILL_DIR}/SKILL.md from disk and resume at the step you are on.

    line 20
  3. Distinctive:

    Use when asked to "resume", "restore context", "where was I", or "pick up where I left off".

    line 24
  4. Distinctive:

    HARD GATE: Do NOT implement code changes.

    line 56
  5. Distinctive:

    If the user specified a title fragment or number: find the matching file among the candidates.

    line 192
  6. Very specific:

    Run this against the chosen file, replacing <CHECKPOINT_FILE> with its path from Step 1.

    line 205
  7. Very specific:

    Every other item goes under Next steps: (path run) with a successful outcome, (path read), (target state checked), and unmarked items that …

    line 260
  8. Distinctive:

    Read the chosen file and present a summary:

    line 269
  9. Very specific:

    If the loaded file has no remote: field, add one line under the header box: "This checkpoint predates project stamps.

    line 304
  10. Very specific:

    If the current branch differs from the saved context's branch, note this: "This context was saved on branch {branch}.

    line 307
  11. Very specific:

    If the loaded file is the current branch's but the NEWEST_OTHER: path from Step 1 (the newest save from another branch) has a newer …

    line 311
  12. Distinctive:

    After presenting, ask via AskUserQuestion:

    line 318
  13. Distinctive:

    If A, take the first Remaining Work item in saved order.

    line 324
  14. Very specific:

    If Step 1 printed NO_CHECKPOINTS, tell the user:

    line 334
  15. Very specific:

    This is a vibestack skill, not a Claude Code built-in: When the user types /context-restore, invoke this skill via the Skill tool.

    line 354

“Common” is measured: kept by at least 6 in 10 of the 18 edited versions of this file we could read. “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

If this session was compacted, the copy of this skill in context may be cut off. Read ${CLAUDE_SKILL_DIR}/SKILL.md from disk and resume at the step you are on.

When to invoke

Use when asked to "resume", "restore context", "where was I", or "pick up where I left off". Pair with /context-save. Formerly /checkpoint resume — renamed because Claude Code treats /checkpoint as a native rewind alias in current environments.

Preamble

eval "$("${VIBESTACK_HOME:-$HOME/.vibestack}/bin/vibe-slug" 2>/dev/null)" 2>/dev/null || SLUG="unknown"
_LEARN_FILE="${VIBESTACK_HOME:-$HOME/.vibestack}/projects/${SLUG:-unknown}/learnings.jsonl"
if [ -f "$_LEARN_FILE" ]; then
  _LEARN_COUNT=$(wc -l < "$_LEARN_FILE" 2>/dev/null | tr -d ' ')
  echo "LEARNINGS: $_LEARN_COUNT entries loaded"
  if [ "$_LEARN_COUNT" -gt 5 ] 2>/dev/null; then
    "${VIBESTACK_HOME:-$HOME/.vibestack}/bin/vibe-learnings-search" --limit 5 2>/dev/null || true
  fi
else
  echo "LEARNINGS: none yet"
fi

{{include lib/snippets/session-host.md}}

{{include lib/snippets/decision-brief.md}}

{{include lib/snippets/working-protocols.md}}

{{include lib/snippets/state-protocols.md}}

/context-restore — Restore Saved Working Context

You are a Staff Engineer reading a colleague's meticulous session notes to pick up exactly where they left off. Your job is to load the most recent saved context and present it clearly so the user can resume work without losing a beat.

HARD GATE: Do NOT implement code changes. This skill only reads saved context files and presents the summary.

Default: load the most recent context saved on the CURRENT branch; if this branch has none, fall back to the most recent across ALL branches. Every worktree of a repo shares one checkpoints directory, so without the preference a sibling worktree's newer save would be presented as "where you left off". The fallback keeps Conductor workspace handoff working — a context saved on one branch can still be resumed from another.

Do NOT hard-filter the candidate set to the current branch. Other-branch files stay in the set; they are ordered after the current branch's own. (/context-save list is the flow that scopes to one branch.)


Detect command

Parse the user's input:

  • /context-restore → load the most recent saved context (current branch first, then any branch)
  • /context-restore <title-fragment-or-number> → load a specific saved context
  • /context-restore list → tell the user "Use /context-save list — listing lives on the save side" and exit. No mode detection here.

Restore flow

Step 1: Find saved contexts

eval "$("${VIBESTACK_HOME:-$HOME/.vibestack}/bin/vibe-slug" 2>/dev/null)" && mkdir -p "${VIBESTACK_HOME:-$HOME/.vibestack}/projects/$SLUG"
CHECKPOINT_DIR="${VIBESTACK_HOME:-$HOME/.vibestack}/projects/$SLUG/checkpoints"
if [ ! -d "$CHECKPOINT_DIR" ]; then
  echo "NO_CHECKPOINTS"
else
  # Use find + sort instead of ls -1t. Two reasons:
  # 1. Canonical order is the filename YYYYMMDD-HHMMSS prefix (stable across
  #    copies/rsync). Filesystem mtime drifts and is not authoritative.
  # 2. On macOS, `find ... | xargs ls -1t` with zero results falls back to
  #    listing cwd. `sort -r` on empty input cleanly returns nothing.
  # Scan the 200 newest so a current-branch save sitting below a burst of
  # sibling-worktree saves is still found; the printed list is capped at 20.
  ALL=$(find "$CHECKPOINT_DIR" -maxdepth 1 -name "*.md" -type f 2>/dev/null | sort -r | head -200)
  # Drop checkpoints stamped for another project (a different remote, or a
  # different root when neither has one). Each is named, never loaded.
  if [ -n "$ALL" ]; then
    CLASSIFIED=$(printf '%s\n' "$ALL" | "${VIBESTACK_HOME:-$HOME/.vibestack}/bin/vibe-slug" --classify-checkpoints 2>/dev/null)
    TAB=$(printf '\t')
    if [ -n "$CLASSIFIED" ] && ! printf '%s\n' "$CLASSIFIED" | grep -qvE "^(match|unstamped|foreign)${TAB}"; then
      printf '%s\n' "$CLASSIFIED" | awk -F'\t' '$(1) == "foreign" { print $(2) }' | while IFS= read -r f; do
        echo "PROJECT MISMATCH: $f ($(grep -m1 '^remote:' "$f" 2>/dev/null), $(grep -m1 '^project_root:' "$f" 2>/dev/null))"
      done
      ALL=$(printf '%s\n' "$CLASSIFIED" | awk -F'\t' '$(1) != "foreign" { print $(2) }')
    else
      echo "IDENTITY_CHECK_UNAVAILABLE"
    fi
  fi
  if [ -z "$ALL" ]; then
    echo "NO_CHECKPOINTS"
  else
    # Current-branch files first, other branches after, each newest first.
    # CURRENT_BRANCH may be preset; otherwise it comes from git.
    : "${CURRENT_BRANCH:=$(git rev-parse --abbrev-ref HEAD 2>/dev/null)}"
    SAME=""; OTHER=""
    while IFS= read -r f; do
      [ -n "$f" ] || continue
      b=$(grep -m1 '^branch:' "$f" 2>/dev/null | sed 's/^branch:[[:space:]]*//')
      if [ -n "$CURRENT_BRANCH" ] && [ "$b" = "$CURRENT_BRANCH" ]; then
        SAME="${SAME}${f}
"
      else
        OTHER="${OTHER}${f}
"
      fi
    done <<EOF
$ALL
EOF
    echo "CURRENT_BRANCH=${CURRENT_BRANCH:-unknown}"
    [ -n "$SAME" ] || echo "NO_CURRENT_BRANCH_CHECKPOINT"
    # Named separately: with 20+ saves on this branch the capped list below
    # would never reach another branch's newer save.
    [ -n "$SAME" ] && [ -n "$OTHER" ] && echo "NEWEST_OTHER: $(printf '%s' "$OTHER" | head -1)"
    # Cap at 20: a user with 10k saved files shouldn't blow the context window
    # just listing them. /context-save list handles pagination.
    printf '%s%s' "$SAME" "$OTHER" | grep -v '^[[:space:]]*$' | head -20
  fi
fi

Candidates include every .md file in the directory, ordered current-branch-first (the branch is read from each file's branch: frontmatter). Other-branch files stay in the set as the fallback, which keeps Conductor workspace handoff working when this branch has no save of its own. NO_CURRENT_BRANCH_CHECKPOINT means every path printed came from another branch.

Project identity. /context-save stamps each checkpoint with remote: (the credential-free origin) and project_root:. A checkpoint stamped for a different project is never a candidate: Step 1 prints PROJECT MISMATCH: <path> (...) for it instead. Report every PROJECT MISMATCH line to the user before any summary, e.g. "Skipped 1 checkpoint saved for github.com/bob/api — it belongs to another project." If that leaves no candidate (NO_CHECKPOINTS), say the bucket holds only other projects' saves. Never load a mismatched file, even when the user names it by title or number — name its project and stop. Checkpoints with no remote: field were saved before stamping existed; they stay candidates. IDENTITY_CHECK_UNAVAILABLE means the installed vibe-slug cannot classify checkpoints — say identity was not checked and suggest /vibe-upgrade.

Auto snapshots. In Claude Code a PreCompact hook writes a git-state snapshot to compact-snapshots/ before every compaction. List the ones newer than the newest saved context:

eval "$("${VIBESTACK_HOME:-$HOME/.vibestack}/bin/vibe-slug" 2>/dev/null)"
SNAP_DIR="${VIBESTACK_HOME:-$HOME/.vibestack}/projects/${SLUG:-unknown}/compact-snapshots"
NEWEST=$(find "${VIBESTACK_HOME:-$HOME/.vibestack}/projects/${SLUG:-unknown}/checkpoints" -maxdepth 1 -name "*.md" -type f 2>/dev/null | sort -r | head -1)
if [ -d "$SNAP_DIR" ]; then
  if [ -n "$NEWEST" ]; then
    find "$SNAP_DIR" -maxdepth 1 -name "*.md" -type f -newer "$NEWEST" 2>/dev/null | sort -r | head -10 | sed 's/^/AUTO_SNAPSHOT: /'
  else
    find "$SNAP_DIR" -maxdepth 1 -name "*.md" -type f 2>/dev/null | sort -r | head -10 | sed 's/^/AUTO_SNAPSHOT: /'
  fi
fi

Show each AUTO_SNAPSHOT: path as "auto snapshot (git state only)" — it is never a candidate for Step 2. A snapshot holds branch, status, recent log and the transcript path — no decisions and no remaining work — so it never replaces a saved context; it only tells the user that work continued after the last save. When no saved context exists, offer the newest snapshot's git state as the only record.

Step 2: Load the right file

  • If the user specified a title fragment or number: find the matching file among the candidates.
  • Otherwise: load the first path printed by Step 1 — the newest YYYYMMDD-HHMMSS save on the current branch or, when Step 1 printed NO_CURRENT_BRANCH_CHECKPOINT, the newest across all branches.
  • No usable branch (NO_CURRENT_BRANCH_CHECKPOINT, or CURRENT_BRANCH is HEAD or unknown): when the newest candidates' titles differ, they are different pieces of work and "newest" is a guess, so ask with AskUserQuestion which one to load (newest first, up to four). When the titles match, it is one task handed across branches; load the newest.

Head check

Run this against the chosen file, replacing <CHECKPOINT_FILE> with its path from Step 1. It compares the commit the save recorded with the commit checked out now:

CK_FILE='<CHECKPOINT_FILE>'
fm() { awk -v k="${1}" 'NR == 1 && $(0) != "---" { exit } NR > 1 && $(0) == "---" { exit }
  NR > 1 && index($(0), k ":") == 1 { v = substr($(0), length(k) + 2); gsub(/^[[:space:]]+|[[:space:]]+$/, "", v); print v; exit }' "$CK_FILE"; }
SAVED=$(fm head); SAVED_WT=$(fm worktree)
NOW=$(git rev-parse --verify -q HEAD 2>/dev/null)
if [ -z "$SAVED" ]; then
  echo "HEAD_UNTRACKED"
elif ! printf '%s' "$SAVED" | grep -qE '^[0-9a-f]{7,64}$' || [ -z "$NOW" ] \
     || ! git cat-file -e "$SAVED^{commit}" 2>/dev/null; then
  echo "HEAD_UNKNOWN"
elif [ "$SAVED" = "$NOW" ]; then
  echo "HEAD_UNCHANGED"
elif git merge-base --is-ancestor "$SAVED" HEAD 2>/dev/null; then
  echo "HEAD_MOVED $(git rev-list --count "$SAVED..HEAD")"
  git log --oneline -20 "$SAVED..HEAD"
else
  echo "HEAD_DIVERGED"
  git log --oneline -20 "$SAVED..HEAD"
fi
if [ -n "$SAVED_WT" ] && [ "$SAVED_WT" != "$(git rev-parse --show-toplevel 2>/dev/null)" ]; then
  echo "SAVED_IN_WORKTREE: $SAVED_WT"
fi
  • HEAD_UNCHANGED — the tree is where the save left it; sort normally.
  • HEAD_MOVED <n>, HEAD_DIVERGED (a rebase, reset or other branch: the saved commit is not behind this one) and HEAD_UNKNOWN (the saved commit is not in this clone): show the listed commits under the header box and put every Remaining Work item under Verify first — anything the save called done or safe may since have been done, undone or changed.
  • HEAD_UNTRACKED — saved before head stamps existed; sort normally.
  • SAVED_IN_WORKTREE: <path> — the save came from another worktree of this repository; name it.

Sort Remaining Work by provenance. /context-save ends each item with a marker saying how that session knew it. Keep every item's original text and saved order, and drop nothing. An item's marker is the last provenance marker in it, and it counts as the ending even when an outcome follows it: older saves wrote Open. Run the suite. (path run) exit 0, current ones write Open. Run the suite, exit 0. (path run), and both are (path run) with a successful outcome. Markers are English literals; a parenthesised token in their place that is none of the five (misspelled, or translated into the session's language) reads as (path assumed). Put an item under Verify first when it:

  • ends in (path assumed) or (code read);
  • ends in (path run) but its text reports a failure;
  • has no marker and is a writing step (migration, sync, insert, import, deploy, a dialog that writes) or names a concrete path (a runnable command, CLI flag or switch, config key or value, or file or directory path).

Every other item goes under Next steps: (path run) with a successful outcome, (path read), (target state checked), and unmarked items that neither write nor name a concrete path. Verifying means read-only inspection: read the file or the target, or run a command that changes nothing. Saves written before provenance markers existed have none, so their concrete-path and writing items land under Verify first. When the file has no provenance markers at all, print this line above the groups: This checkpoint predates provenance markers; items naming commands, paths or writes are listed under Verify first.

Read the chosen file and present a summary:

RESUMING CONTEXT
════════════════════════════════════════
Title:       {title}
Branch:      {branch from frontmatter}
Saved:       {timestamp, human-readable}
Duration:    Last session was {formatted duration} (if available)
Status:      {status}
Head:        {head check result}
════════════════════════════════════════

### Summary
{summary from saved file}

### Next Focus
{Next Focus from the saved file, when present}

### Remaining Work
{legacy banner line, if it applies}

Next steps
{Next steps items, in saved order, original text}

Verify first (inspect read-only before executing anything)
{Verify first items, in saved order, original text}

### Notes
{notes}

### Suggested skills / References
{as saved, when present}

If the loaded file has no remote: field, add one line under the header box: "This checkpoint predates project stamps; confirm it belongs to this project."

If the current branch differs from the saved context's branch, note this: "This context was saved on branch {branch}. You are currently on {current branch}. You may want to switch branches before continuing."

If the loaded file is the current branch's but the NEWEST_OTHER: path from Step 1 (the newest save from another branch) has a newer timestamp, name it: "There is also a newer context saved on {other branch} — {title} from {timestamp}. Load that one instead?" A Conductor handoff to this branch shows up that way, and the user decides.

Step 3: Offer next steps

After presenting, ask via AskUserQuestion:

  • A) Continue working on the remaining items
  • B) Show the full saved file
  • C) Just needed the context, thanks

If A, take the first Remaining Work item in saved order. If it is under Next steps, suggest starting there. If it is under Verify first, suggest verifying it read-only before doing it or any later item, so a later runnable step never jumps ahead of an unverified earlier one. Never execute a Verify first item as part of the restore.


If no saved contexts exist

If Step 1 printed NO_CHECKPOINTS, tell the user:

"No saved contexts yet. Run /context-save first to save your current working state, then /context-restore will find it."


Important Rules

  • Never modify code. This skill only reads saved files and presents them.
  • Prefer the current branch's own save, but keep all branches in the fallback set. Cross-branch resume (Conductor handoff) works when this branch has no save, and a sibling worktree's newer save never shadows this branch's.
  • Another project's checkpoint is never restored. A PROJECT MISMATCH file is reported and skipped, never presented as "where you left off".
  • Unverified steps are shown, not run. An item under Verify first was guessed or failed in the saved session; inspect before acting on it.
  • "Most recent" means the filename YYYYMMDD-HHMMSS prefix, not ls -1t (filesystem mtime). Filenames are stable across file-system operations; mtime is not.
  • This is a vibestack skill, not a Claude Code built-in. When the user types /context-restore, invoke this skill via the Skill tool.

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.