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-step-execution- SKILL.md3.8 KB
Overview
This skill enforces precise execution of a single step from a specification and returns an unambiguous result. It is used whenever an agent must perform one step literally, detect gaps, run the step, verify outcomes, and report SUCCESS, FAILED, or GAP. The skill is strict about scope: no extra changes, no judgment calls, and immediate gap reporting.
How this skill works
The skill reads the spec and locates the target step, noting the step description, expected outcome, and verification items. It runs a gap test before and during execution; any ambiguity, missing prerequisites, or required judgment triggers an immediate GAP result. If no gap is found, it executes only the explicit instructions, completes verifications, and returns the result in the required structured format (SUCCESS, FAILED, or GAP).
When to use it
- Executing one explicit step from a formal spec with required auditability
- Automated agents performing incremental work where each step must be validated
- Workflows that demand strict traceability and unambiguous outcomes
- Situations where making judgment calls is unacceptable without spec changes
- When verification items are part of the step and must be enforced
Best practices
- Always supply spec file path, step ID/name, and any progress or research files before starting
- Perform the gap test thoroughly and stop immediately if any ambiguity arises
- Execute only the actions described in the step; avoid improvements or side effects
- Record verification results explicitly and mark the step FAILED if any item fails
- Return the exact result block format specified and include artifacts modified when relevant
Example use cases
- Run a single test-case step in a test plan and verify the listed assertions
- Apply a specific code change described verbatim in a development spec
- Collect a single research item from a research spec and validate source citations
- Perform a one-step data-cleaning instruction and confirm the verification checks
- Execute a deployment step that contains explicit commands and post-checks
FAQ
Any situation where the spec lacks necessary detail, requires interpretation, has unmet prerequisites, or encounters unexpected conditions not covered by the spec.
Can I improve or optimize the step during execution?
No. The skill requires following the spec literally. Suggested improvements should be recorded as observations or proposed spec edits, but not implemented.