Use when starting or resuming work in a workspace that keeps passdown handoff logs — reads the latest handoff and plan state, verifies them against the working tree, and briefs the session so work resumes from files instead of a replayed transcript
Scanned 8/30/2026
Install to Claude Code
npx -y skills add vunm-io/passdown --skill passdown-pickup --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Passdown Pickup?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/vunm-io-passdown-pickup)More formats (shields.io, HTML) on the badges page.
---
name: passdown-pickup
description: Use when starting or resuming work in a workspace that keeps passdown handoff logs — reads the latest handoff and plan state, verifies them against the working tree, and briefs the session so work resumes from files instead of a replayed transcript
---
# Passdown Pickup
Start of shift: read the previous passdown before touching anything. The
handoff wrote state into small files; pickup turns those files back into
working context for a fraction of what replaying a transcript costs.
## Configuration (read first)
Build the effective passdown configuration root-to-nearest. Read applicable
`AGENTS.md` files from the workspace/repository root down to the current
directory, plus any parent file explicitly referenced by a thin entrypoint.
Merge `## passdown` keys in that order: nearer values override the same key and
inherit omitted keys. Resolve relative paths against the file that declared
them.
The effective configuration should contain:
```markdown
## passdown
- log_dir: <path for session logs>
- planning: markdown | openspec # used to locate plan state
- plan_dir: <path for markdown plans> # when planning: markdown
```
If `log_dir` is still missing, ask where session logs live and suggest adding
it to the appropriate `AGENTS.md`.
## Process
1. **Locate the handoff**: list `<log_dir>` and sort by the filename's date
and `HHMMSS` suffix, newest first. Read the newest log's frontmatter
(`status`, `branch`, `agent`, `plan`); also read the frontmatter of any
other recent logs, in case another agent worked in parallel. Frontmatter
first, full text only for the logs that matter — that is what keeps pickup
cheap.
2. **Read the passdown**: for the relevant log(s), read Summary, Next steps,
and especially Caveats / traps in full. Task state lives in the plan;
session-scoped traps live nowhere but the log.
3. **Verify against reality** — files may have moved on since the log was
written:
- the current branch vs the log's `branch:`;
- `git status` and recent commits vs the log's "What was done";
- each plan listed in `plan:` (a single path or a YAML list, resolved
from the workspace root) — compare its checkboxes and any
`Dispatched:` outcome lines against the log's next steps.
Report any mismatch instead of silently reconciling it; the working tree
and the plan win over the log.
4. **Brief, then wait**: report status, verified next steps, and the traps.
Pickup produces a briefing and a proposed first action — it
never starts executing on its own. If the proposed work is a multi-task
plan, the `passdown-dispatch` gate still applies before implementation.
## Rules
- An empty or missing `log_dir` is a fact to report, not an error to fix:
say so and brief from plan/task state alone.
- Never modify a previous shift's log. Corrections and discoveries belong in
this session's own handoff or the workspace inbox.
- An `IN_PROGRESS` or `BLOCKED` log from another agent may mean that shift is
still active — surface it before building on that work.
- Trust files over the transcript: when a summarized conversation and the
logs disagree, the logs and the plan are the source of truth.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!