Health Gated Task Finisher
A structured execution framework for agents featuring health-gated loops, automated rollbacks, and state resumption.
Free
Works with the AI tools you already use
Claude Code
Cursor
Codex CLI
Muse
OpenClaw+21 moreClaude CodeCursorCodex CLIMuseOpenClaw
Manus
Grok
Grok Bot+18 agentsHealth Gated Task Finisher
Example session with this skill installed
Run a standard task to refactor the payment gateway. The goal is to move stripe-specific logic to a separate provider file. Health is GREEN, confidence is HIGH. Use L0 channel.
- Read your context and instructions
- Compiled the health gated task
I have initiated the Health-Gated Task Finisher loop for the payment gateway refactor. The task has been classified as
Standard due to multiple file mutations and state dependency.
Execution Summary
| Phase | Status | Result |
|---|---|---|
| Distill | PASS | Contract established: Decouple Stripe logic from PaymentService.ts. |
| Break Apart | PASS | 4 steps defined (S01-S04). S04 marked as hard reversibility. |
| Execute (S01) | PASS | Created providers/StripeProvider.ts. Verification: File exists and exports interface. |
| Execute (S02) | PASS | Migrated createCharge logic. Verification: Unit test stripe-prov.test.ts passed. |
| Integrate | PASS | Updated PaymentService.ts imports. Publication safety check cleared. |
Task State
{
"task_id": "refactor-pay-2023",
"status": "COMPLETED",
"health": "GREEN",
"streak": 2,
"last_channel": "L0",
"artifacts": [
"src/providers/StripeProvider.ts",
"src/services/PaymentService.ts"
],
"verification": {
"type": "unit-test",
"result": "PASS",
"evidence": "tests/logs/refactor-output.log"
}
}
Next steps
- Review the generated
StripeProvider.tsfor any missing edge cases in the error handling block. - Run the full integration suite to ensure the DI container correctly resolves the new provider.
- Commit the changes if the
STATEfile shows no pending rollbacks or unresolved RED signals.
Connects securely to your tools. The creator never sees your data.
About this skill
The problem
Agentic workflows often fail silently when state degrades or tasks unexpectedly drift into irreversible territory. Developers struggle with agents that hallucinate completion or lack the safety rails to rollback failed operations during complex multi-step executions.
What it does
- Enforces a 5-phase execution loop (Distill, Break, Execute, Integrate, Hand Off) with mandatory pre-gate and post-verify beats.
- Implements a "Two-Signal Rule" requiring high confidence and green health status before any autonomous write.
- Manages task escalation via a tiered channel system (L0-L3) to reroute work based on risk and reversibility.
- Handles automatic recovery and rollback using pre-action snapshots and a strictly defined retry budget.
- Maintains a canonical STATE file to allow a different agent or human to resume work exactly where it left off.
Why this beats prompting it yourself
Standard prompting relies on the LLM's "memory" which fades or corrupts over long sequences. This skill replaces vague instructions with a structured state machine and an append-only health log, ensuring zero-bypass safety checks that a simple prompt cannot sustain across multiple turns.
Use cases
- Executing multi-file refactors where each step must be verified before the next begins.
- Managing deployment scripts or database migrations that require explicit compensation plans for failure.
- Orchestrating hand-offs between different agents in a pipeline without losing context or intent.
- Running background maintenance tasks that must halt and escalate if environment health degrades.
Known limitations
Assumes reliable file I/O for state persistence. Does not support network or browser tools; limited to project and spool file scopes.
How to install
Works the same in every agent - Claude, Cursor, Codex, Copilot and 20+ more.
- 1
Download the ZIP
Free skills download straight away. Paid skills unlock right after purchase.
- 2
Unzip into your skills folder
Every agent reads skills from one folder on your machine. Drop the unzipped folder in there.
- 3
Ask your agent to use it
Restart the agent if it was already running. It picks the skill up automatically - no config needed.
Skills folder by agent
Click the path to copy it. Create the folder if it does not exist yet.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill passes our trust checks below.
Security scanned
Passed our 8-point scan before listing
2 installs
Downloaded by developers to date
Free forever
No account required to browse
Trust & safety
Security scanned
Verified clean 17 days ago
- Free to download with an account