agentleFS
IlyaGorskyIlyaGorsky/memory-toolkit/skills/session-continue/SKILL.md

session-continue

Skill13 starsChanged 11 days agoMITView on GitHub

Continue a workstream from where you left off — faster than /session-start, no focus selection.

The file

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 7 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

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

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

    If no handoff exists → suggest /session-start instead.

    line 30
  2. Distinctive:

    If there are uncommitted changes — show in one line.

    line 42
  3. Distinctive:

    If the tree is clean — don't mention it.

    line 43
  4. Very specific:

    Don't show task plan, focus candidates, model recommendations — that's for /session-start.

    line 61
  5. Distinctive:

    If the user confirmed — load only what's needed for the next step:

    line 69
  6. Distinctive:

    Do NOT load the entire memory — only what's relevant.

    line 75
  7. Distinctive:

    Don't ask about the model, don't show candidates — the user already knows

    line 90
  8. Distinctive:

    If handoff is older than 7 days — warn: "Handoff from {date}, context may be stale.

    line 91
  9. Distinctive:

    If no argument is passed — take the workstream from the latest handoff

    line 92

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

/session-continue — Quick Continue

Pick up a workstream without a full cold start. For cases when you know what you're doing.


Step 1: Find handoff

node "${CLAUDE_PLUGIN_ROOT}/scripts/memory.js" handoff "$ARGUMENTS"

Falls back to global handoff if no per-workstream file exists. Prints NO_HANDOFF when neither found.

node "${CLAUDE_PLUGIN_ROOT}/scripts/memory.js" card "$ARGUMENTS"

The workstream's card — what it is, what's open. Read it silently to frame the handoff; NO_CARD is fine.

If no handoff exists → suggest /session-start instead.


Step 2: Git status (silently)

git status --short
git log --oneline -3
git branch --show-current

If there are uncommitted changes — show in one line. If the tree is clean — don't mention it.


Step 3: Show summary

Format (10 lines max):

## Continuing: {workstream}

**Where we left off:** {from handoff — last action}
**What's next:** {from handoff — next step}
**Branch:** {branch} {uncommitted count if any}

Shall we begin?

Don't show task plan, focus candidates, model recommendations — that's for /session-start.


Step 4: Load context on demand

MUST wait for explicit user confirmation ("yes", "go", a number, or similar) before proceeding. Do NOT auto-continue after showing the summary.

If the user confirmed — load only what's needed for the next step:

  • The file that was last edited
  • Task plan if available
  • Last 3 feedback entries via node "${CLAUDE_PLUGIN_ROOT}/scripts/memory.js" search

Do NOT load the entire memory — only what's relevant.


Step 5: Mark

node "${CLAUDE_PLUGIN_ROOT}/scripts/memory.js" note "SESSION_CONTINUE branch:$(git branch --show-current) workstream:{name}"

Rules

  • Maximum speed — 10 lines of output, then get to work
  • Don't ask about the model, don't show candidates — the user already knows
  • If handoff is older than 7 days — warn: "Handoff from {date}, context may be stale. /session-start for a full overview?"
  • If no argument is passed — take the workstream from the latest handoff

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.