Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add iroha924/mumei/plugin install mumeiWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/iroha924/mumei/shelve)<a href="https://agentmods.dev/skills/iroha924/mumei/shelve"><img src="https://agentmods.dev/badge/skills/iroha924/mumei/shelve/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/iroha924/mumei/shelve"><img src="https://agentmods.dev/badge/skills/iroha924/mumei/shelve.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00095 | $0.01942 |
| Opus 5 | $0.00048 | $0.00971 |
| Sonnet 5 | $0.00019 | $0.00388 |
| Haiku 4.5 | $0.00010 | $0.00194 |
Grade A, and why
shelve scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 153 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Shelve
Move a completed feature out of the active workspace into the archive directory. This skill is user-invocable only (disable-model-invocation: true) — Claude will not auto-trigger shelving even if the workflow seems "done".
When to use
- The user explicitly invokes
/mumei:shelve <feature>. - A feature has
phase: doneand the user is ready to clean up the active workspace.
Pre-flight checks
Refuse with a clear error if any of these fail:
<feature>slug must exist as a directory under either.mumei/specs/(spec vehicle) or.mumei/plans/(plan vehicle). Try specs/ first, then plans/. Refuse if neither has the slug.state.jsonmust havephase: "done"(orphase: "review"with the latest review verdictPASS, with explicit confirmation).- Working tree must be clean for files within the feature's
_Files:_scope (spec vehicle only — plan vehicle has no_Files:_meta and skips this check). .mumei/currentis exclusively owned by this skill. No other skill or hook may clear it. If<feature>is the active feature in.mumei/current, this skill auto-clears the file as part of the shelve operation (see Method below). The "owned exclusively" rule prevents session-handoff inconsistency where a prior turn cleared.mumei/currentwhile leaving the spec / plan dir behind, causing the next session to lose track of in-progress work.
Method
source "${CLAUDE_PLUGIN_ROOT}/hooks/_lib/state.sh"
feature="$1"
# auto-detect vehicle by directory existence. spec vehicle
# (.mumei/specs/) takes precedence when both happen to exist; the slug
# collision picker in /mumei:compose is supposed to prevent that situation
# in the first place.
source_dir=""
state_path=""
if [[ -d ".mumei/specs/${feature}" ]]; then
source_dir=".mumei/specs/${feature}"
state_path=".mumei/specs/${feature}/state.json"
elif [[ -d ".mumei/plans/${feature}" ]]; then
source_dir=".mumei/plans/${feature}"
state_path=".mumei/plans/${feature}/state.json"
else
echo "Feature not found: ${feature} (looked in .mumei/specs/ and .mumei/plans/)" >&2
exit 1
fi
# Both vehicles store phase in the same field; mumei_state_read_any
# returns the value from whichever state.json exists.
phase="$(mumei_state_read_any "$feature" '.phase' 2>/dev/null || true)"
if [[ "$phase" != "done" ]]; then
echo "Feature ${feature} is not done (phase=${phase}). Refuse." >&2
exit 1
fi
# Phase D — cross-feature dependency guard.
# Refuse to shelve when an active feature declares a Wave-level
# `**Depends-Feature**:` directive pointing at this feature. The user
# can override by either shelving the dependency (remove the
# directive in the dependent's tasks.md) or by archiving in the
# correct order (dependents first).
source "${CLAUDE_PLUGIN_ROOT}/hooks/_lib/dependencies.sh"
dependents="$(mumei_dependencies_active_dependents_of "$feature" 2>/dev/null || true)"
if [[ -n "$dependents" ]]; then
echo "Cannot shelve ${feature}: active dependent feature(s) still declare it via Wave **Depends-Feature**:" >&2
printf ' %s\n' $dependents >&2
echo "Either shelve the dependents first, or remove the Depends-Feature line." >&2
exit 1
fi
# Calculate archive subdir based on creation month (or current month if missing).
# Both schemas have created_at as ISO 8601.
created_at="$(mumei_state_read_any "$feature" '.created_at' 2>/dev/null || true)"
yyyymm="$(date -u -d "${created_at}" +%Y-%m 2>/dev/null || date -u +%Y-%m)"
target_dir=".mumei/archive/${yyyymm}"
mkdir -p "$target_dir"
# Refuse if target already exists (collision)
if [[ -e "${target_dir}/${feature}" ]]; then
echo "Archive target already exists: ${target_dir}/${feature}" >&2
exit 1
fi
# Capture the glean scratch path BEFORE moving the source dir
# (mumei_state_read_any no-ops once the dir is moved). Prefer the recorded
# scratch_source — it survives a feature slug that diverged from the scratch
# basename (collision -N suffix, rename), so the originating scratch is
# co-moved instead of orphaned. Fall back to the legacy slug-match for
# features predating the field. When neither resolves to an existing file,
# the co-move below silently skips.
slug="$(mumei_state_read_any "$feature" '.slug' 2>/dev/null || true)"
[[ -z "$slug" ]] && slug="$feature"
scratch_src="$(mumei_state_scratch_source "$feature" 2>/dev/null || true)"
[[ -z "$scratch_src" ]] && scratch_src=".mumei/scratch/${slug}.md"
# Move the source directory. The move + git history serves as
# the audit trail. Refuse to continue if both git mv and the bare mv
# fallback fail — without an explicit guard the scratch block would
# still run on a half-shelved feature.
git mv "$source_dir" "${target_dir}/${feature}" 2>/dev/null \
|| mv "$source_dir" "${target_dir}/${feature}" \
|| { echo "source dir move failed: ${source_dir}" >&2; exit 1; }
# Move the glean scratch file alongside the spec / plan, if
# present. Vehicle-independent: plan vehicle uses the same scratch
# co-move behaviour as spec vehicle.
if [[ -n "$scratch_src" && -f "$scratch_src" ]]; then
scratch_dst="${target_dir}/${feature}/scratch.md"
git mv "$scratch_src" "$scratch_dst" 2>/dev/null \
|| mv "$scratch_src" "$scratch_dst" \
|| mumei_log_warn "shelve: scratch co-move failed for ${scratch_src} (left in place); move it manually or it will linger in .mumei/scratch/"
fi
# Auto-clear .mumei/current if it points at the feature being shelved.
if [[ -f .mumei/current ]]; then
current="$(tr -d '[:space:]' <.mumei/current)"
if [[ "$current" == "$feature" ]]; then
: >.mumei/current
fi
fi
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 153 lines · 95 tokens per session scan A ad7c200d8561
shelve is a skill published in the GitHub repository iroha924/mumei (2 stars, last pushed 5d ago), licensed MIT. It adds 95 tokens to every session and 1,942 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
md-hive-sync
Munder Difflin hive sync — runs the start-of-task hive protocol steps: reads memory.md, checks inbox/ for new messages, and reminds you to record durable facts in memory.md and write coordination files before ending. Use when asked to "sync with the hive", "check my inbox", "hive status", or "hive sync". Proactively…
harness
A meta-skill for designing and maintaining a project harness: a coordinated set of specialist agents, skills, and project instructions. It defines their roles, connects them to the project, and updates the setup as the work changes.
create-request
Create, update, or scan per-task request tickets for progress tracking. These are date-prefixed non-lifecycle docs under requests/, NOT feature-level requirements (use /req-analyze for those). Use when: tracking task progress, updating completion status, scanning incomplete requests, checking request status dashboard.…
req-analyze
A requirements-analysis tool that breaks a problem into needs, stakeholders, and priorities before a technical solution is chosen.
recap-ask
Interactive Q&A over an existing recap document. Use when: user wants to ask follow-up questions about a briefing-recap- .md produced by /recap-doc, with recap-bounded context + out-of-scope redirect + optional promote-to-request. Not for: generating a new recap (use /recap-doc), general project Q&A (use /ask), code…
recap-doc
Post-development recap document generator. Use when: AI/Codex has implemented a feature and the user needs a guided walkthrough of what changed and why, with blind-spot detection and anticipated questions. Not for: Q&A follow-up (use /recap-ask), technical share-out for teammates (use /tech-brief), or generic code…