1
GitHub Stars
1
Bundled Files
3 weeks ago
Catalog Refreshed
2 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstart where the catalogue uses aiagentskills.
npx veilstart add skill wilddeer/specops --skill spec-execution- SKILL.md12.1 KB
Overview
This skill executes existing specifications literally and without creativity. It follows the spec step-by-step, documents progress immediately, and stops to flag any gaps. When a gap is found, it switches to spec-development to resolve it and then resumes execution.
How this skill works
Before doing any work, I ask which spec to execute and locate the spec file. I run each step exactly as written: execute, spawn a reviewer, handle the review result, commit only after approval, and update the progress file immediately after each step. If the spec defines parallel execution, I launch subagents per the plan; if missing details are encountered, I pause and hand off to spec-development. Final completion requires a final review, verification checklist, and explicit user confirmation.
When to use it
- You have a completed spec and want deterministic, literal execution.
- You want strict audit trails and per-step progress logging.
- You need reproducible runs that avoid agent creativity or assumptions.
- You must preserve spec authority and only change specs via controlled development.
- You require orchestration that can run steps sequentially or via approved subagents.
Best practices
- Always confirm which spec file and execution mode (direct or subagents) before starting.
- Create or confirm a progress file location before executing the first step.
- Show proposed changes, pause for user confirmation, then execute — one change per message.
- Update the progress file immediately after each step with status, results, files changed, and blockers.
- When reviewers return APPROVED_WITH_RECOMMENDATIONS or GAP, pause and route to spec-development before continuing.
Example use cases
- Running a deployment procedure specified in a runbook where no interpretation is allowed.
- Executing a QA verification checklist that must be logged step-by-step.
- Performing a multi-step migration that includes sequential and parallel parts defined in the spec.
- Automating compliance checks where every action must match the written spec exactly.
- Orchestrating tests that require external reviewers to approve each change before committing.
FAQ
I stop and flag the gap, then switch to spec-development to fill the missing details before resuming execution.
Do you make changes without user approval?
No. I always present proposed changes, show the exact modification, ask for confirmation, and only act after the user decides.