Install
$ agentstack add skill-kks0488-vibe-codex-vc-phase-loop ✓ 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
VC Phase Loop
Outcome
Complete the requested work within the user's scope and authority, then support the completion claim with executed checks and concrete evidence.
1. Scope and risk
- Read applicable repository instructions before editing.
- Separate requirements from assumptions.
- Identify destructive, costly, credential-sensitive, public, or externally consequential actions.
- Proceed with reversible in-scope choices. Ask when an unanswered choice materially changes the
result or requires authority the user has not granted.
2. Choose task depth
- Small: execute directly, then verify.
- Multi-step: keep a short plan with one active step.
- Long-running or resumable: create
.vc/work-{timestamp}.mdwith requirements, decisions, status,
blockers, and completion evidence. Update it at decision or phase boundaries, not after every command.
Do not inflate a task to reach a fixed phase count.
3. Execute
- Follow existing project patterns.
- Keep changes minimal and reviewable.
- Use built-in Codex subagents only for independent work that benefits from separate context, such as
exploration, test execution, or review.
- Avoid parallel write-heavy work unless file ownership is clearly separated.
- Preserve user changes and stay within the approved roots.
4. Recover intelligently
Use the Two-Strike rule:
- Diagnose the first failure and try a targeted correction.
- If the same failure repeats, stop using the same approach.
- Re-check assumptions, inspect the root cause, and choose a materially different strategy.
- If progress requires missing authority, credentials, money, user input, or external state, report
the blocker instead of pretending to continue.
Retries are bounded by new information. Never promise guaranteed success or an infinite loop.
5. Verify proportionally
Prefer the project's own checks in this order when available:
- focused tests for changed behavior;
- lint, type, or syntax checks;
- broader tests or build for higher-risk changes;
- a real user-path exercise when practical.
Do not claim a check passed unless it ran successfully. If a check could not run, state why.
6. Report
Lead with the outcome. Include:
- what changed;
- commands or user paths actually verified;
- important assumptions or tradeoffs;
- remaining blockers or follow-up work.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: kks0488
- Source: kks0488/vibe-codex
- License: MIT
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.