Skip to content
Back to skills

Execute

ASecurity

Execute one accepted task and leave verifiable evidence.

  • 6 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added September 6, 2026
development

Security analysis

A100/100

Pro scans all 10 files and shows the line behind each finding

Scanned September 6, 2026

npx -y skills add BeforeWave/forma --skill execute --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Execute?

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

Security grade badge for Execute
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/beforewave-execute-forma/badge)](https://www.skillsdirectory.com/skills/beforewave-execute-forma)

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: execute
description: "Execute one accepted task and leave verifiable evidence."
---

# Forma Execute

Execute one accepted task and leave verifiable evidence.

## Interaction Semantics

- Use this skill as a review-gated task executor for one accepted task contract.
- For each task, implement only the current task's accepted files/surfaces, run its validation gates through review-ready, and wait for explicit user approval before completion.
- After an approved task is completed successfully, check for the next task immediately and continue if one remains.
- Stop only when review is needed, no unchecked task remains, validation fails, workflow fails, escalation is needed, or the user interrupts.

## Workflow

- Read `./plans/issue-<id>/plan.md`.
- Use `scripts/forma-workflow.sh next <issue-id>` from this installed skill package to get the current task.
- Resolve `scripts/forma-workflow.sh` relative to the current triggered skill package only; never switch to a same-named script in a sibling skill directory, even if the contents match.
- Implement only the current task by treating `Accept:` as the delivery target and `Validate:` / `Use-Check:` as validation gates and proof obligations; if it is the last remaining task, `review-ready` will also run `## Final Validation`.
- For `loop-batch`, report metric/artifact movement before validation; for `gate`, name the protected decision boundary; for `promote`, write only the surfaces explicitly allowed by the current task.
- Update `plans/issue-<id>/implement-notes.md` before `review-ready` whenever this task made any execution choice, changed the locked contract (`plan.md` or `tasks.md`), deviated from the task scope, or touched files beyond the planned surface. Record the choice, plan gap, classification, deviation, or follow-up with the bundled notes reference. When the task executed exactly as planned with no such event, record `Decision Notes: None beyond the task plan` in the current task section; absence of a current task section is not an acceptable substitute.
- Before review-ready for behavior, architecture, ownership, or integration work, load `references/engineering-quality-model.md`, self-check the applicable dimensions against the final diff and evidence, and resolve or record material findings within the task contract.
- Run `scripts/forma-workflow.sh review-ready <issue-id>` before presenting the current task result for user review.
- `scripts/forma-workflow.sh review-ready <issue-id>` stages the reviewed task snapshot; if review feedback changes the task, rerun `review-ready` before completion.
- Only after the user explicitly approves the current task for completion, run `scripts/forma-workflow.sh complete <issue-id>`.
- After `complete` succeeds, immediately run `scripts/forma-workflow.sh next <issue-id>`; if another unchecked task is returned, start executing it in the same invocation instead of waiting for the user to ask again.

Read these files first:

- `./plans/issue-<id>/plan.md`
- `./plans/issue-<id>/tasks.md`

## Load As Needed

- `references/execution-rules.md`
- `references/implement-notes.md`
- `references/engineering-quality-model.md`
- `references/task-runner.md`

## Requirements

- Load and follow `references/task-runner.md` for review-gated task execution.

## Output

- Follow `references/output-format.md`.

Files in this skill

  • SKILL.md3.3 KB
  • agents/openai.yaml222 B
  • references/engineering-quality-model.md4.4 KB
  • references/execution-rules.md6.9 KB
  • references/implement-notes.md4 KB
  • references/output-format.md2.5 KB
  • references/plan-template.md9.1 KB
  • references/task-runner.md4.8 KB
  • references/tasks-template.md1.7 KB
  • scripts/forma-workflow.sh51.2 KB

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.