app-implement
Implements app based on APP_BRIEF.md delta, supports dry-run capability
How do I install this agent skill?
npx skills add https://github.com/cribl-community/claude-cribl-app-plugins --skill app-implementIs this agent skill safe to install?
- Gen Agent Trust Hubpass
The skill automates the implementation of Cribl applications by processing instructions from a design brief. It incorporates strong security guidelines, such as secret encryption and strict directory scoping. The verdict is low because the skill treats the content of the brief file as an authoritative source of truth, creating a potential surface for indirect prompt injection.
- Socketpass
No alerts
- Snykwarn
Risk: MEDIUM · 1 issue
What does this agent skill do?
Pre-flight Check
Before proceeding, verify you are in a scaffolded Cribl app folder. If you are not, stop and tell the user: "You must run this skill from within a scaffolded Cribl app folder."
Instructions
You are implementing a Cribl app based on an APP_BRIEF.md file. Use Capra UI by default for all components. Before starting, review ./references/cribl-apps-guidance.md to avoid common pitfalls with KV store, state management, error handling, and other Cribl-specific patterns.
Follow this workflow:
1. Validate Prerequisites
- Check that
APP_BRIEF.mdexists in the working directory - If missing, error with: "APP_BRIEF.md not found. Run /app-brief first."
- Create
/executionsdirectory if it doesn't exist
2. Load & Compare Versions
- Read the current
APP_BRIEF.md - Find the most recent
APP_BRIEF_*.mdin/executions(sort by timestamp in filename) - If no prior execution: treat entire current brief as new work (delta = all of it)
- If prior execution exists: diff the two to identify:
- New sections/requirements
- Modified sections/requirements
- Sections removed (note: don't undo these)
3. Parse Delta into Action Items
- Extract specific implementation tasks from the delta
- For each task, create one clear, actionable line describing what will be done
- Example lines:
- "Create new component: UserInput"
- "Update API endpoint /search to support filtering"
- "Add test coverage for AuthService"
4. Dry-Run Mode (if --dry-run flag provided)
- Print header: "=== DRY RUN: Planned Changes ==="
- Print each action item as one line (no descriptions, just the action)
- Print footer: "=== End Dry Run ==="
- Example output:
=== DRY RUN: Planned Changes === Create new component: UserInput Update API endpoint /search to support filtering Add test coverage for AuthService === End Dry Run ===
5. Implementation Mode (normal execution, no --dry-run)
- Execute the delta: create files, modify code, update configs as needed
- Read the app id from APP_DEFINITION.md (under "## App ID" section)
- Update the "name" field in package.json to match the app id
- Remove all linter errors, run
npm run lint
6. Archive This Run
- Copy current
APP_BRIEF.mdto/executions/APP_BRIEF_[ISO8601-timestamp].md - Use format:
APP_BRIEF_2026-07-31T14-30-45Z.md(ISO 8601, with colons replaced by hyphens for filesystem compatibility) - This becomes the baseline for the next run
7. Confirm Package.json Update
- Verify that package.json's "name" field now matches the app id from APP_DEFINITION.md
Key Principles
- Source of truth: Current
APP_BRIEF.mdalways wins - Idempotency: Skill can run repeatedly; each run only touches what changed
- Stateless between runs:
/executionsfolder is the only state persistence
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/cribl-community/claude-cribl-app-plugins/app-implement">View app-implement on skillZs</a>