Sweep the full current session, repo/task context, local task files, and (with approval) connected remote task stores for outstanding actionable work, then record only useful high-confidence items in docs/outstanding-issues.md with priority, context, and next steps. Use when the user types "/ledger" or asks to extract outstanding tasks from this session into the ledger.
89%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
Low
Low-risk findings worth noting
/ledger is a session sweep: it mines everything visible to the current task for outstanding
actionable items and records the high-confidence ones in docs/outstanding-issues.md — the same
universal ledger the /issues skill owns. /ledger is the extractor;
/issues remains the reader/editor. Follow every writing rule in the issues skill.
Disambiguation: this skill targets the outstanding-issues ledger only. It must never read or
write docs/branch-review-ledger.md, and the ledger:lookup / ledger:index npm scripts
belong to that review history — they are not part of this skill.
Collect candidates from, in order:
git status, uncommitted/unpushed work, recent commits
on this branch, and TODO/FIXME markers in files touched this session (not a repo-wide grep).docs/outstanding-issues.md (baseline for dedupe),
docs/process-hardening.md debts, session scratchpad notes, and any plan/handoff notes this
task produced.Hunt specifically for: deferred or "later" items; unresolved "need to", "should", "TODO", "follow up", "please do", "investigate", "fix" statements; explicit blockers and errors left unfixed; and acceptance criteria stated but not met.
Present findings in this order, then pause for confirmation on anything ambiguous:
/ledger itself (report only; add to
the ledger only if the user confirms).Render each candidate with the full structured record:
id: short-slug # e.g. ledger-missing-trigger
title: one line
importance: P0 | P1 | P2 | P3
category: bug | follow-up | risk | refactor | infrastructure | hygiene | idea
why: business value / user impact
context: what in this chat triggered it
owner: assistant | user | unknown
next_step: concrete next action
dependencies: ids or #NNN rows, if any
status: pending | done | blocked
confidence: high | medium | low
suggested_improvements: optional related idea (omit if none)npm run issues:update
instead of a new row.Queue each retained item as an immutable inbox request; never edit the canonical ledger:
npm run issues:add -- --pri P2 --type issue --summary "<title>" --detail "<detail>" --source "session YYYY-MM-DD /ledger sweep"Field mapping into the request payload (the reconciler later creates the canonical row):
importance → Pri: P0 and P1 → P1 (note "was P0" in Detail), P2 → P2, P3 → P3.category → Type: bug/risk → issue; follow-up/refactor/infrastructure/hygiene → task;
idea → rec.next_step, then Why:, Context:, Owner:, Confidence:, and
Depends on: as short clauses. Escape | as \|. Keep it one crisp cell, not a paragraph.session YYYY-MM-DD /ledger sweep plus the doc/PR/file:line when one exists.If no new items survive the filter, create no request and do not change the canonical ledger.
After queueing, run npm run check:outstanding-issues and paste its decisive line. Do not refresh
the visual register because the canonical ledger is unchanged. If the user explicitly asks for a
commit, commit only the newly created request file(s); never push unless the user asks or a handoff
is already in flight. A dedicated fresh-base branch later runs npm run issues:reconcile.
Fallback (other workspaces): if docs/outstanding-issues.md does not exist, search for
similarly named files (outstanding*, *issues*, TODO.md, BACKLOG.md); if none fit, propose
a location and ask before creating. A newly created ledger uses concise structured Markdown
bullets carrying the full record above, appended chronologically, preserving existing entries.
/ledger never deletes or rewrites existing rows — cleanup only on explicit request, via
/issues.BigSimmo/PsychSift
0d5a97a
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.