Install
$ agentstack add skill-upsolve-labs-upstack-upstack-run ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
Update Check (run first)
_UPD=$(~/.claude/skills/upstack/bin/upstack-update-check 2>/dev/null || .claude/skills/upstack/bin/upstack-update-check 2>/dev/null || true)
[ -n "$_UPD" ] && echo "$_UPD" || true
If output shows UPGRADE_AVAILABLE : read ~/.claude/skills/upstack/skills/upgrade/SKILL.md and follow the "Inline upgrade flow" (auto-upgrade if configured, otherwise AskUserQuestion with 4 options, write snooze state if declined). If JUST_UPGRADED : tell user "Running upstack v{to} (just updated!)" and continue.
Upstack Run
Chains the full sprint flow into a single run. The end result is a pushed PR.
Step 1: Plan
Always run /plan. The /plan skill has its own Fast-Path Check that detects when a plan already exists in context and the new request is a small follow-up — in that case it skips full ceremony and produces a lightweight amendment. If the change is complex, /plan runs its full flow. Either way, do NOT proceed until /plan completes.
Step 2: Execute
Run /execute against the approved plan. Strict Red/Green TDD: write failing tests, implement until green, atomic commits.
Step 3: Validate
Run /validate. Walk through every planned path manually, capture evidence (screenshots, API examples) to evidence/.
IMPORTANT: Do NOT skip /validate when UI files are in the diff. "Playwright tests pass" is NOT a substitute for /validate. Tests verify assertions; validate produces evidence (screenshots, API examples) for the PR and catches visual/UX issues that assertions miss.
If validation finds gaps:
- Amend the plan with the gaps.
- Loop back to Step 2 (/execute) to close them.
- Re-validate. Maximum 3 loops before stopping and reporting to the user.
Step 4: Review
Run /review. Senior engineer code review against the base branch.
If review finds CRITICAL or SECURITY findings:
- Automatically fix the flagged issues (treat them as if the user said "fix all").
- Loop back to Step 2 (/execute) for the fixes.
- Re-run /review. Maximum 3 review/execute loops before stopping and reporting to the user.
Step 5: Ship (MANDATORY — do not skip)
Run /ship-pr. This step MUST execute — it handles documentation, version bump, and PR creation. All three are required outputs of /upstack-run. The version bump question is the only interactive pause. If you have reached this step, do not end the conversation without completing /ship-pr.
Guardrails
- Max loops: 3 for validate gaps, 3 for review findings. If still failing after 3, stop and report what remains.
- Abort on hard failure: If tests cannot pass, gh is unavailable, or the branch is in a bad state, stop and report. Do not retry endlessly.
- No silent skips: Always announce which step you are entering and whether any steps were skipped or looped.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: Upsolve-Labs
- Source: Upsolve-Labs/upstack
- License: MIT
- Homepage: https://upsolve.ai/
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.