Skip to content
Back to skills

Innovate Loopback

ASecurity

Run the Vibe Innovation Framework mini-gate assessment to scope, justify, and execute a loop-back. Use when evidence in the current phase suggests an earlier phase's output is wrong, when a red-team challenge cannot be addressed in the current phase, or when an assumption from an earlier phase has been falsified.

  • 4 stars
  • 0 votes
  • 0 copies
  • 3 views
  • Added September 12, 2026
ai-agentsgo

Security analysis

A100/100

Pro scans all 2 files and shows the line behind each finding

Scanned September 12, 2026

npx -y skills add aibot88/sec_skill_store --skill innovate-loopback --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Innovate Loopback?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Innovate Loopback
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/aibot88-innovate-loopback/badge)](https://www.skillsdirectory.com/skills/aibot88-innovate-loopback)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: innovate-loopback
description: Run the Vibe Innovation Framework mini-gate assessment to scope, justify, and execute a loop-back. Use when evidence in the current phase suggests an earlier phase's output is wrong, when a red-team challenge cannot be addressed in the current phase, or when an assumption from an earlier phase has been falsified.
user-invocable: true
argument-hint: "[trigger evidence]"
---

**Loop-back trigger:** $ARGUMENTS

Read the following files in order before acting.

1. `.claude/docs/loop_back_protocol.md` (the canonical runbook with the eight-step mini-gate)
2. `.claude/docs/orchestrator.md` (the *Loop-back and iteration protocol* section, for budget rules and escalation options)
3. `.claude/docs/principles_and_antipatterns.md` (Principle 8: iteration has a budget)
4. The current ICD Section 7 (Iteration log) and the relevant earlier phase prompt only after the target phase is identified.

Then execute the runbook exactly as specified there. The runbook is self-contained: trigger confirmation, type and target identification, scope, carry-forward inventory, budget check, re-entry pre-amble, forward propagation, and logging.

## Operating notes

1. **Budget enforcement is non-negotiable.** Maximum 2 intra-phase iterations, maximum 2 inter-phase loop-backs to the same target, maximum 5 total inter-phase loop-backs. The kill threshold rule (any assumption falsified in two separate cycles is dead) applies regardless of remaining budget.
2. **Pause, do not reset.** The current phase's output is preserved and marked *provisional*. The phase resumes after the loop-back resolves. Do not delete the in-progress work.
3. **Re-entry pre-amble is verbatim.** When the user runs `/innovate-phase N` for the target phase, the Step 1 pre-amble must include all five fields from runbook Step 6. Do not paraphrase.
4. **Logging is mandatory.** Append a row to ICD Section 7 with the loop type, from-phase, to-phase, trigger evidence, scope, and outcome. Update iteration counters. Add a Section 9 (Changelog) entry with the same date.
5. **Escalation handling.** If a budget limit is reached, do not loop back. Surface the four escalation options (accept lower TRL, grant one more iteration with specific evidence expected, pivot, or kill) and let the user decide. Do not pre-empt the decision.
6. **Working language.** Detect the user's working language and translate the eight runbook prompts. Keep the structure faithful and the prompts terse.

Respond in the language the user is writing in.

Files in this skill

  • SKILL.md2.5 KB
  • source.json833 B

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.