How long does the integration take?

Change one base URL in an OpenAI-compatible client, or install the Python or TypeScript SDK; the quickstart routes a first request in a few minutes.

Will this lock us into another vendor?

No. The router is Apache-2.0 open source, your provider keys stay yours, and the API is OpenAI-compatible, so leaving is a change of base URL.

What happens when a provider fails?

The router falls back to the next model your policy allows and records the fallback in the trace, so an outage costs a retry, not an incident.

triage - Prompt - OpenSmartRoute

Name: triage
Author: jupes

Prompt file imported from jupes/agent-forge-harness (.claude/commands/triage.md). Copyright stays with the author.

/triage — Deadline Management & Capacity Planning

Classifies epics by deadline risk, reviews open PRs across repos, and creates bugs for failing CI.

Usage

/triage
/triage --epic EPIC-7        # Triage a single epic

Step 1 — Load All Epics

bd ready --type epic 2>/dev/null
bd ready --status in_progress --type epic 2>/dev/null

For each epic: bd show <EPIC-ID> to get due date, task count, completed count.

Step 2 — Classify Each Epic

For each epic, calculate:

Total tasks: child task count
Completed tasks: closed child task count
Remaining tasks: open + in-progress child tasks
Days until due: (due date) - (today)
Velocity needed: remaining tasks / days until due

Classify:

Status	Criteria
OVERDUE	Due date has passed and epic is not closed
AT RISK	Due in <7 days with >30% tasks remaining, OR blocked tasks on critical path
ON TRACK	Due in ≥7 days with ≤30% remaining, OR no due date
COMPLETE	All tasks closed

Step 3 — Review Open PRs (all repos)

gh pr list --state open --json number,title,headRefName,statusCheckRollup,reviews,createdAt --limit 20

For each open PR, check:

CI status: PASS / FAIL / PENDING
Review status: approved / changes requested / awaiting
Age: how many days old

Flag PRs that need attention:

CI failing → create Beads bug (see Step 4)
No review after 2+ days → flag for reviewer assignment
Changes requested but no author response after 1+ days → flag

Step 4 — Auto-Create Bugs for Failing PRs

For each PR with failing CI:

bd create \
  --type bug \
  --title "[<repo>] CI failing on PR #<n>: <pr title>" \
  --priority 1 \
  --acceptance "CI passes on PR #<n>"

bd comments add <new-bug-id> "design: CI failure detected by /triage. PR: #<n>. Failures: <ci check names>"

Priority policy: keep high for merge-blocking CI on important branches; use critical only when production or security is directly at risk. When re-ranking work, use .claude/skills/beads-priority-assignment/SKILL.md.

Only create if a bug for this PR doesn't already exist:

bd ready | grep -i "PR #<n>"   # Skip if already filed

Step 5 — Capacity Summary

List all ready tasks by priority to help with planning:

bd ready | head -20

Count by priority:

Critical: N tasks
High: N tasks
Medium: N tasks
Low: N tasks

Identify any tasks blocked for more than 3 days — flag them as potential abandonment candidates.

Output Format

## Triage Report — <date>

### Epics

🔴 OVERDUE (1)
  EPIC-3  "User permissions system"  due 2026-04-05  (6 days overdue)
           8/12 tasks complete  →  4 tasks remain

🟠 AT RISK (1)
  EPIC-5  "Dashboard redesign"  due 2026-04-18  (7 days)
           3/10 tasks complete  →  7 tasks remain  (needs ~1/day)
           ⚠ T-67 blocked (dependency on T-54)

🟢 ON TRACK (2)
  EPIC-7  "Settings page"  due 2026-04-25  (14 days)
           5/8 tasks complete  →  3 tasks remain
  EPIC-9  "API rate limiting"  no due date
           1/4 tasks complete

---

### PR Health (4 open)

  #17  feat(auth): refresh tokens         CI: PASS   ✓ Approved    (2 days old)
  #18  fix(dash): pagination              CI: FAIL   ← created T-71 bug
  #16  chore: update deps                 CI: PENDING (in progress)
  #15  feat(settings): user prefs         CI: PASS   ✗ Awaiting review (3 days)

  Action: Assign reviewer to #15

---

### Capacity (ready tasks)

  Critical: 1    High: 5    Medium: 8    Low: 3

  Longest-blocked:
  T-49  "Migrate to v2 API"  blocked 5 days  → consider deferring

---

### Actions Created
  T-71 [bug/high]  CI failing on PR #18: fix dashboard pagination

Install triage and it becomes one more candidate the router can pick - when it fits.

A listing is a routing target with a manifest: what it does, which domains it covers, what it costs and who publishes it. Once installed it sits beside your own models and tools, is scored like any other candidate for each request, and shows up in the trace when it wins. Ratings come from workspaces that installed it, one per account.

# /triage — Deadline Management & Capacity Planning Classifies epics by deadline risk, reviews open PRs across repos, and creates bugs for failing CI. ## Usage ``` /triage /triage --epic EPIC-7 # Triage a single epic ``` --- ## Step 1 — Load All Epics ```bash bd ready --type epic 2>/dev/null bd ready --status in_progress --type epic 2>/dev/null ``` For each epic: `bd show <EPIC-ID>` to get due date, task count, completed count. --- ## Step 2 — Classify Each Epic For each epic, calculate: - **Total tasks**: child task count - **Completed tasks**: closed child task count - **Remaining tasks**: open + in-progress child tasks - **Days until due**: (due date) - (today) - **Velocity needed**: remaining tasks / days until due Classify: | Status | Criteria | |--------|----------| | **OVERDUE** | Due date has passed and epic is not closed | | **AT RISK** | Due in <7 days with >30% tasks remaining, OR blocked tasks on critical path | | **ON TRACK** | Due in ≥7 days with ≤30% remaining, OR no due date | | **COMPLETE** | All tasks closed | --- ## Step 3 — Review Open PRs (all repos) ```bash gh pr list --state open --json number,title,headRefName,statusCheckRollup,reviews,createdAt --limit 20 ``` For each open PR, check: - **CI status**: PASS / FAIL / PENDING - **Review status**: approved / changes requested / awaiting - **Age**: how many days old Flag PRs that need attention: - CI failing → create Beads bug (see Step 4) - No review after 2+ days → flag for reviewer assignment - Changes requested but no author response after 1+ days → flag --- ## Step 4 — Auto-Create Bugs for Failing PRs For each PR with failing CI: ```bash bd create \ --type bug \ --title "[<repo>] CI failing on PR #<n>: <pr title>" \ --priority 1 \ --acceptance "CI passes on PR #<n>" bd comments add <new-bug-id> "design: CI failure detected by /triage. PR: #<n>. Failures: <ci check names>" ``` **Priority policy:** keep **`high`** for merge-blocking CI on important branches; use **`critical`** only when production or security is directly at risk. When re-ranking work, use `.claude/skills/beads-priority-assignment/SKILL.md`. Only create if a bug for this PR doesn't already exist: ```bash bd ready | grep -i "PR #<n>" # Skip if already filed ``` --- ## Step 5 — Capacity Summary List all ready tasks by priority to help with planning: ```bash bd ready | head -20 ``` Count by priority: - Critical: N tasks - High: N tasks - Medium: N tasks - Low: N tasks Identify any tasks blocked for more than 3 days — flag them as potential abandonment candidates. --- ## Output Format ``` ## Triage Report — <date> ### Epics 🔴 OVERDUE (1) EPIC-3 "User permissions system" due 2026-04-05 (6 days overdue) 8/12 tasks complete → 4 tasks remain 🟠 AT RISK (1) EPIC-5 "Dashboard redesign" due 2026-04-18 (7 days) 3/10 tasks complete → 7 tasks remain (needs ~1/day) ⚠ T-67 blocked (dependency on T-54) 🟢 ON TRACK (2) EPIC-7 "Settings page" due 2026-04-25 (14 days) 5/8 tasks complete → 3 tasks remain EPIC-9 "API rate limiting" no due date 1/4 tasks complete --- ### PR Health (4 open) #17 feat(auth): refresh tokens CI: PASS ✓ Approved (2 days old) #18 fix(dash): pagination CI: FAIL ← created T-71 bug #16 chore: update deps CI: PENDING (in progress) #15 feat(settings): user prefs CI: PASS ✗ Awaiting review (3 days) Action: Assign reviewer to #15 --- ### Capacity (ready tasks) Critical: 1 High: 5 Medium: 8 Low: 3 Longest-blocked: T-49 "Migrate to v2 API" blocked 5 days → consider deferring --- ### Actions Created T-71 [bug/high] CI failing on PR #18: fix dashboard pagination ```

{ "prompt": "# /triage — Deadline Management & Capacity Planning\n\nClassifies epics by deadline risk, reviews open PRs across repos, and creates bugs for failing CI.\n\n## Usage\n```\n/triage\n/triage --epic EPIC-7 # Triage a single epic\n```\n\n---\n\n## Step 1 — Load All Epics\n\n```bash\nbd ready --type epic 2>/dev/null\nbd ready --status in_progress --type epic 2>/dev/null\n```\n\nFor each epic: `bd show <EPIC-ID>` to get due date, task count, completed count.\n\n---\n\n## Step 2 — Classify Each Epic\n\nFor each epic, calculate:\n- **Total tasks**: child task count\n- **Completed tasks**: closed child task count\n- **Remaining tasks**: open + in-progress child tasks\n- **Days until due**: (due date) - (today)\n- **Velocity needed**: remaining tasks / days until due\n\nClassify:\n\n| Status | Criteria |\n|--------|----------|\n| **OVERDUE** | Due date has passed and epic is not closed |\n| **AT RISK** | Due in <7 days with >30% tasks remaining, OR blocked tasks on critical path |\n| **ON TRACK** | Due in ≥7 days with ≤30% remaining, OR no due date |\n| **COMPLETE** | All tasks closed |\n\n---\n\n## Step 3 — Review Open PRs (all repos)\n\n```bash\ngh pr list --state open --json number,title,headRefName,statusCheckRollup,reviews,createdAt --limit 20\n```\n\nFor each open PR, check:\n- **CI status**: PASS / FAIL / PENDING\n- **Review status**: approved / changes requested / awaiting\n- **Age**: how many days old\n\nFlag PRs that need attention:\n- CI failing → create Beads bug (see Step 4)\n- No review after 2+ days → flag for reviewer assignment\n- Changes requested but no author response after 1+ days → flag\n\n---\n\n## Step 4 — Auto-Create Bugs for Failing PRs\n\nFor each PR with failing CI:\n\n```bash\nbd create \\\n --type bug \\\n --title \"[<repo>] CI failing on PR #<n>: <pr title>\" \\\n --priority 1 \\\n --acceptance \"CI passes on PR #<n>\"\n\nbd comments add <new-bug-id> \"design: CI failure detected by /triage. PR: #<n>. Failures: <ci check names>\"\n```\n\n**Priority policy:** keep **`high`** for merge-blocking CI on important branches; use **`critical`** only when production or security is directly at risk. When re-ranking work, use `.claude/skills/beads-priority-assignment/SKILL.md`.\n\nOnly create if a bug for this PR doesn't already exist:\n```bash\nbd ready | grep -i \"PR #<n>\" # Skip if already filed\n```\n\n---\n\n## Step 5 — Capacity Summary\n\nList all ready tasks by priority to help with planning:\n\n```bash\nbd ready | head -20\n```\n\nCount by priority:\n- Critical: N tasks\n- High: N tasks\n- Medium: N tasks\n- Low: N tasks\n\nIdentify any tasks blocked for more than 3 days — flag them as potential abandonment candidates.\n\n---\n\n## Output Format\n\n```\n## Triage Report — <date>\n\n### Epics\n\n🔴 OVERDUE (1)\n EPIC-3 \"User permissions system\" due 2026-04-05 (6 days overdue)\n 8/12 tasks complete → 4 tasks remain\n\n🟠 AT RISK (1)\n EPIC-5 \"Dashboard redesign\" due 2026-04-18 (7 days)\n 3/10 tasks complete → 7 tasks remain (needs ~1/day)\n ⚠ T-67 blocked (dependency on T-54)\n\n🟢 ON TRACK (2)\n EPIC-7 \"Settings page\" due 2026-04-25 (14 days)\n 5/8 tasks complete → 3 tasks remain\n EPIC-9 \"API rate limiting\" no due date\n 1/4 tasks complete\n\n---\n\n### PR Health (4 open)\n\n #17 feat(auth): refresh tokens CI: PASS ✓ Approved (2 days old)\n #18 fix(dash): pagination CI: FAIL ← created T-71 bug\n #16 chore: update deps CI: PENDING (in progress)\n #15 feat(settings): user prefs CI: PASS ✗ Awaiting review (3 days)\n\n Action: Assign reviewer to #15\n\n---\n\n### Capacity (ready tasks)\n\n Critical: 1 High: 5 Medium: 8 Low: 3\n\n Longest-blocked:\n T-49 \"Migrate to v2 API\" blocked 5 days → consider deferring\n\n---\n\n### Actions Created\n T-71 [bug/high] CI failing on PR #18: fix dashboard pagination\n```", "variables": [], "notes": "Classifies epics by deadline risk, reviews open PRs across repos, and creates bugs for failing CI.", "metadata": { "source": { "provider": "github-claude-commands", "repository": "https://github.com/jupes/agent-forge-harness", "path": ".claude/commands/triage.md", "ref": "e47bc8ad9842926586c567103e90bb8c0f32a543", "url": "https://github.com/jupes/agent-forge-harness/blob/e47bc8ad9842926586c567103e90bb8c0f32a543/.claude/commands/triage.md", "key": "jupes/agent-forge-harness/.claude/commands/triage.md" } } }

triage

About

/triage — Deadline Management & Capacity Planning

Usage

Step 1 — Load All Epics

Step 2 — Classify Each Epic

Step 3 — Review Open PRs (all repos)

Step 4 — Auto-Create Bugs for Failing PRs

Step 5 — Capacity Summary

Output Format

Install triage and it becomes one more candidate the router can pick - when it fits.

What it costs to adopt - in engineering time.

One base URL

No vendor lock-in

Failover built in

Use it

Manifest

Reviews