| name | wr-itil:work-problem |
| description | Pick the highest-WSJF open or known-error problem ticket and work it — investigate, implement, commit, and release per the standard manage-problem workflow. Selection is framework-mediated (WSJF + documented tie-break ladder per ADR-044); singular variant; distinct from /wr-itil:work-problems (plural AFK orchestrator). Use this when the user asks to "work the next problem", "work the top of the queue", or "grind through one ticket". |
| allowed-tools | Read, Write, Edit, Bash, Glob, Grep, AskUserQuestion, Skill, Agent |
Work Problem — Pick-and-Run
Pick the highest-WSJF ticket from the current backlog and work it end-to-end. This is the singular entry point — one ticket per invocation. Selection is framework-mediated per ADR-044's Framework-Mediated Surface (Prioritisation row): the agent applies the ADR-076 tier partition (Critical-bypass → Inbound-reported → Internal) then the WSJF formula + documented within-tier tie-break ladder mechanically, and reports the chosen ticket + its tier + the rung that decided. The user retains direct override via /wr-itil:work-problem <NNN> invocation and via mid-flow correction (ADR-044 category 6).
This skill is the P071 phased-landing split of /wr-itil:manage-problem work per ADR-010 amended Skill Granularity rule: one skill per distinct user intent. The original /wr-itil:manage-problem work subcommand route remains as a thin-router forwarder during the deprecation window but is scheduled for removal in @windyroad/itil's next major version.
Name distinction (work-problem vs work-problems)
/wr-itil:work-problem (singular, this skill) — one ticket per invocation. Framework-mediated selection (WSJF + tie-break ladder). Intended for a user who wants to dispatch the next-highest ticket and then stop. User-override path: /wr-itil:work-problem <NNN> to pin a specific ticket./wr-itil:work-problems (plural, AFK orchestrator) — loops through the backlog by WSJF, delegating each iteration to this skill (via the Agent tool, per ADR-032 + P077). Intended for AFK batch runs; non-interactive selection; stops only when nothing actionable remains.
Both names coexist intentionally per P071's out-of-scope note on the naming coexistence. The plural orchestrator uses this skill as its per-iteration unit.
Scope
In scope:
- Read
docs/problems/README.md when fresh; otherwise delegate the refresh to /wr-itil:review-problems first (never re-implement the re-scoring logic locally — same anti-fork discipline as the list-problems cache path).
- Pick the highest-priority ticket via the framework-mediated selection: first by ADR-076 tier (Tier 0 Critical-bypass [Severity Very High ≥17 OR security-classified OR incident-linked] → Tier 1 Inbound-reported [
**Origin**: inbound-reported] → Tier 2 Internal), then within the highest non-empty tier by the tie-break ladder (Known Error > Open; smaller effort first; older reported date; ticket number ascending) per ADR-044 Framework-Mediated Surface. Report the chosen ticket + its tier + the rung that decided. No AskUserQuestion fires for selection.
- Honour the user-override path
/wr-itil:work-problem <NNN> — when the user names a ticket directly, skip the ladder and proceed to Step 3 with the named ticket.
- Delegate the actual work to
/wr-itil:manage-problem <NNN> via the Skill tool so the investigation / known-error transition / fix / closure flow stays hosted on a single authoritative workflow (ADR-010 thin-router discipline applied to the work path).
- Scope-expansion prompt (ADR-013 Rule 1; ADR-044 category-2 deviation-approval surface) when the selected ticket's effort grows during work — same three-option structure as
/wr-itil:manage-problem's Working a Problem section.
Out of scope:
- Batch looping over multiple tickets — that's
/wr-itil:work-problems (plural). This skill runs exactly one ticket per invocation.
- Backlog re-ranking / Verification Queue prompt / README.md refresh — that's
/wr-itil:review-problems. This skill reads the ranking; it does not rewrite it.
- Ticket creation — that's
/wr-itil:manage-problem (no-args form).
- Status transitions other than those driven by the delegated
/wr-itil:manage-problem <NNN> flow.
Steps
1. Ensure the ranking is fresh
Check the docs/problems/README.md cache with the same git-history-based freshness test as /wr-itil:list-problems and /wr-itil:review-problems (per P031 — filesystem mtime is unreliable in worktrees and fresh checkouts, so git history is the authoritative staleness signal):
readme_commit=$(git log -1 --format=%H -- docs/problems/README.md 2>/dev/null)
if [ -z "$readme_commit" ] || \
git log --oneline "${readme_commit}..HEAD" -- 'docs/problems/*.md' 'docs/problems/*/*.md' ':!docs/problems/README.md' 2>/dev/null | grep -q .; then
echo "stale"
fi
- Cache fresh (no output): read
docs/problems/README.md and use the cached WSJF Rankings table for Step 2.
- Cache stale (prints "stale") or
README.md missing: delegate to /wr-itil:review-problems via the Skill tool to refresh the ranking before proceeding. Do NOT re-implement the re-scoring logic here — that would fork the review path and break P062's canonical-cache-writer contract. The review skill's Step 4 verification prompt runs on this refresh path (P048 Candidate 1: Verification Queue prompts always fire so pending verifications don't accumulate off-ledger).
2. Select the ticket (framework-mediated)
Read the WSJF Rankings table from the now-fresh docs/problems/README.md. Apply the framework's tie-break ladder mechanically to pick the next ticket — selection is framework-mediated per ADR-044's Framework-Mediated Surface (Prioritisation row). The agent picks, reports the choice + the tie-break rung that decided, and proceeds. No AskUserQuestion fires for selection — the WSJF formula + tie-break ladder already resolve the decision.
User-override path — when the user invokes /wr-itil:work-problem <NNN> (ticket ID supplied as an argument, e.g. /wr-itil:work-problem 042), skip the ladder entirely and proceed to Step 3 with the named ticket. This is the documented escape hatch for the user to bypass the framework-mediated selection. Mid-flow correction (ADR-044 category 6 / P078 surface) is the long-tail catcher when the agent's pick was wrong in a way the framework couldn't anticipate.
Tie-break ladder (applied in order; same logic the plural orchestrator's Step 3 uses, per P077 + ADR-044):
- WSJF score (descending) — highest WSJF wins.
- Status — Known Error before Open. Verification Pending tickets are not in this queue (handled by
/wr-itil:review-problems Step 4 verification prompt).
- Effort — smaller effort first (S < M < L < XL).
- Reported date — older reported date wins (FIFO discipline for tied tickets at this rung).
- Ticket number — ascending (final deterministic break).
When a single ticket is the strict top, rungs 2-5 are not consulted. When multiple tickets are tied at the top, walk down the ladder until a rung decides.
Report shape (JTBD-201 audit-trail) — after picking, the agent reports the chosen ticket + the rung that decided, with citations to the WSJF inputs. Example: "Selected P042: (WSJF 12.0; tied with P057 at WSJF; tie-break rung: 'Known Error > Open' — P042 is Known Error, P057 is Open)". This citation makes the selection reproducible from the README state at the time of the report and is the audit surface JTBD-201 requires.
No prose-ask fallback — never present the selection as prose "(a)/(b)/(c)" or "which would you like?" (regression guard for ADR-013 Confirmation grep). Selection is mechanical; the agent picks and reports.
No AFK / interactive split — selection is mechanical in both interactive and AFK modes. The prior asymmetry between AFK (mechanical) and interactive (AskUserQuestion-driven) was the lazy-deferral surface ADR-044 was written to close; converging both modes to the same algorithm is the alignment. The plural orchestrator (/wr-itil:work-problems) Step 3 also uses this ladder, so all entry surfaces resolve to one selection algorithm (no AFK-vs-interactive divergence to maintain).
3. Delegate the work to /wr-itil:manage-problem <NNN>
Invoke /wr-itil:manage-problem <NNN> via the Skill tool with the selected ticket's ID as the argument. The delegated skill runs the full Working a Problem flow appropriate to the ticket's status:
- Open ticket: investigate root cause; document findings; create reproduction test; identify workaround; auto-transition to Known Error when root cause + workaround are documented; if the fix is small, proceed straight into implementation.
- Known Error: traverse problem
## Fix Strategy → referenced RFCs → each RFC's frontmatter stories: array (ordered per ADR-060 line 259) → pick first not-done story (status accepted or in-progress, skipping done). Per ADR-096 a draft story is NOT silently skipped: if the first not-done story is draft, refuse-and-route — surface accept STORY-<NNN> via /wr-itil:manage-story <NNN> accepted first (it runs the INVEST + RFC-trace gates + ratification) and (under AFK) queue it to outstanding_questions rather than skipping past it or implementing it (the itil-no-implement-draft-gate blocks the implementing commit regardless) → implement the picked story scope. An empty stories: [] is a legacy/back-fill state (per ADR-089 every RFC has ≥1 story — the empty-stories atomic fallback is removed): back-fill a story onto the RFC, don't fall back to a per-RFC atomic dispatch. No-RFC Fix Strategy (Phase 1-shape legacy problems) falls back to direct fix implementation. Each commit carries Refs: STORY-<NNN> (or Refs: RFC-<NNN> for cross-cutting RFC work with no single story / Refs: P<NNN> for legacy direct path). Stories auto-transition accepted → in-progress on the first implementing commit (draft → in-progress REMOVED per ADR-096 — a draft story is blocked from implementation); auto-transition in-progress → done on all-criteria-ticked + linked RFC closed. When all stories under all referenced RFCs are done, include the Known Error → Verification Pending git mv in the final commit per ADR-022. The full traversal contract lives in /wr-itil:manage-problem § Working a Problem → Known Error subsection (post-Phase-2 rewrite per ADR-060 lines 300-320).
Why delegate rather than re-implement: the full investigation / transition / fix / release pipeline is a long-lived, policy-governed flow that must stay on a single authoritative workflow. Re-hosting it on a sibling skill would fork the ownership contract and compound maintenance cost. The split skill (this file) owns the selection of the next ticket; /wr-itil:manage-problem <NNN> owns the execution.
4. Scope-expansion check (ADR-013 Rule 1; ADR-044 category-2 deviation-approval)
If the delegated /wr-itil:manage-problem <NNN> reports that the ticket's effort expanded during investigation or architect review (e.g. S → L, or L → XL), fire the standard scope-change AskUserQuestion prompt. Effort growth IS the contradicting evidence against the WSJF score that ranked this ticket at the top — this is the ADR-044 category-2 (deviation-approval) surface. The 3-option vocabulary below is the work-item-tactical analog of the framework-tactical 5-option vocabulary (Approve+amend / Approve+supersede / Approve+one-time / Reject / Defer); the user is the right authority for the shape, so the AskUserQuestion is genuine, not lazy.
- Option 1:
Continue with expanded scope — keep working this ticket at its new size.
- Option 2:
Update problem and re-rank — save findings to the problem file, re-score WSJF, and re-run the framework-mediated selection to surface the new top of queue.
- Option 3:
Pick a different problem — park this one and work something else.
Use header: "Scope change" and multiSelect: false. This is the same structure /wr-itil:manage-problem's Working a Problem section documents; the split lifts it without modification.
AFK / non-interactive branch (ADR-013 Rule 6): same fallback as /wr-itil:work-problems — save findings + the new Effort/WSJF lines to the ticket file and skip to the next iteration. The orchestrator owns the re-selection loop.
5. Post-work housekeeping
After the delegated /wr-itil:manage-problem <NNN> completes:
- If the ticket status changed (Open → Known Error, Known Error → Verification Pending), the delegated skill already committed the change per ADR-014 + ADR-022 and updated
docs/problems/README.md per P062. This skill does NOT re-commit or re-refresh; it reports what happened.
- Report the outcome: ticket ID, action taken (investigated / transitioned / fix-released), commit SHA(s), and next-recommended action (typically "run
/wr-itil:work-problem again for the next ticket" or "stop — all actionable tickets worked").
- Do NOT loop automatically. This is the singular skill. If the user wants to continue, they invoke
/wr-itil:work-problem again, or switch to /wr-itil:work-problems (plural) for the AFK orchestrator loop.
Ownership boundary
work-problem (singular) owns:
- Reading the WSJF Rankings from
docs/problems/README.md.
- Applying the framework-mediated tie-break ladder for selection (ADR-044 Prioritisation row).
- Reporting the chosen ticket + the tie-break rung that decided (JTBD-201 audit-trail).
- Honouring the user-override path
/wr-itil:work-problem <NNN> (skip the ladder; proceed with the named ticket).
- Delegating one execution iteration to
/wr-itil:manage-problem <NNN>.
- Firing the scope-expansion
AskUserQuestion on effort drift (ADR-044 category-2 deviation-approval surface).
work-problem does NOT:
- Fire
AskUserQuestion for the ticket selection itself — selection is framework-mediated per ADR-044.
- Refresh the README.md cache directly — defers to
/wr-itil:review-problems.
- Loop over multiple tickets — that's
/wr-itil:work-problems (plural).
- Commit the ranking refresh or the per-ticket work — delegated skills commit per ADR-014.
Goal anchor for headless runs (P390 / ADR-094)
A headless single-ticket run can anchor its completion with Claude Code's native /goal external evaluator (≥ v2.1.139), so a fresh model — not the working agent — judges whether the ticket genuinely reached an end state: claude -p "/goal Run /wr-itil:work-problem to work the top ticket. Complete when the report printed in the conversation shows a committed outcome (with commit SHA) or a recorded blocker for the selected ticket." (No turn-bound: trust the goal — the loop stops only at its real end states: a committed outcome, a recorded blocker, or quota exhaustion.) There is no programmatic mid-session surface for setting a goal (probed 2026-07-06, v2.1.201) — interactive users type /goal themselves; the skill proceeds identically either way. The plural orchestrator's anchor contract (canonical condition, printed-evidence rule, one-directional semantics) lives at /wr-itil:work-problems Step 0e.
Related
- P071 (
docs/problems/071-argument-based-skill-subcommands-are-not-discoverable.open.md) — originating ticket. This skill is phase 3 of the P071 phased-landing plan (list-problems was phase 1; review-problems was phase 2).
- P136 (
docs/problems/136-adr-044-alignment-audit-master.open.md) — ADR-044 alignment audit master; this skill is the Phase 2 first audit target (work-problem singular).
- ADR-010 amended (
docs/decisions/010-rename-wr-problem-to-wr-itil.proposed.md — Skill Granularity section) — canonical skill-split naming + forwarder contract + deprecated-arguments: true frontmatter flag.
- ADR-013 amended (
docs/decisions/013-structured-user-interaction-for-governance-decisions.proposed.md) — Rule 1 amended in P135 to defer to ADR-044 for framework-resolution boundary. Step 4 scope-expansion is the only Rule-1 surface this skill retains; Rule 6 covers Step 4's AFK fallback.
- ADR-044 (
docs/decisions/044-decision-delegation-contract.proposed.md) — Decision-Delegation Contract; this skill's Step 2 selection is framework-mediated per the ADR's Prioritisation row. Step 4 scope-expansion is a category-2 (deviation-approval) surface per the ADR's 6-class taxonomy.
- ADR-014 — governance skills commit their own work. The delegated
/wr-itil:manage-problem <NNN> owns the per-ticket commit; this skill does not re-commit.
- ADR-018 — release cadence. AFK orchestrator owns release cadence; this skill does NOT auto-release.
- ADR-032 — governance skill invocation patterns.
/wr-itil:work-problems delegates iterations via the Agent tool; this singular skill is the canonical execution unit.
- ADR-037 (
docs/decisions/037-skill-testing-strategy.proposed.md) — contract-assertion bats pattern applied to this skill.
- P031 — git-history freshness check rationale (mtime unreliable in worktrees). Applies to the README cache this skill reads.
- P062 —
/wr-itil:review-problems is the canonical README.md cache writer. This skill defers to it for refreshes.
- P077 —
/wr-itil:work-problems Step 5 delegates iterations via the Agent tool. The delegated subagent invokes this skill's execution unit per iteration.
- JTBD-001 (
docs/jtbd/developer/JTBD-001-enforce-governance.proposed.md) — discoverable surface via /wr-itil: autocomplete. Users type /wr-itil:work-problem rather than remembering the manage-problem work subcommand.
- JTBD-101 (
docs/jtbd/plugin-developer/JTBD-101-extend-suite.proposed.md) — one skill per distinct user intent.
packages/itil/skills/manage-problem/SKILL.md — hosts the thin-router forwarder for the deprecated manage-problem work form; also the delegated execution target for each ticket.packages/itil/skills/review-problems/SKILL.md — sibling refresh skill; this skill defers to it when the cache is stale.packages/itil/skills/list-problems/SKILL.md — sibling read-only display skill; same cache read contract.packages/itil/skills/work-problems/SKILL.md — plural AFK orchestrator; uses this skill as its per-iteration execution unit.
$ARGUMENTS
