OpenSpec Implement Skill

When to Invoke

Automatically activate when user:

• Says "Implement this change", "Apply the fix", "Execute proposal"

• Asks "Apply this bug fix", "Implement the proposal", "Execute the change"

• Has an approved OpenSpec proposal ready to implement

• Mentions "implement", "apply", "execute" with Level 0-1 context

• Uses words like: apply, execute, implement, change, fix, proposal

Specific trigger phrases:

• "Implement this change"

• "Apply the bug fix"

• "Execute the proposal"

• "Apply this change: [proposal-id]"

• "Implement proposal [X]"

• "Execute the fix"

Prerequisites:

• OpenSpec proposal exists and is approved

• Change is still Level 0-1 (hasn't grown in scope)

• Environment and dependencies are clear

Do NOT invoke when:

• No proposal exists (use openspec-change-proposal first)

• Scope has grown beyond Level 1 (escalate to BMAD)

• Implementing a BMAD story (use bmad-development-execution instead)

Auto-escalate to BMAD when:

• Implementation reveals hidden complexity

• Scope expands beyond original proposal

• Tests fail repeatedly indicating design issues

Mission

Apply small code or configuration changes approved via OpenSpec proposals, ensuring each task is executed transparently with testing evidence.

Inputs Required

• proposal: latest proposal.md with decision history

• tasks: tasks.md with sequenced work items and owners

• environment: information about repositories, branches, and tooling needed for execution

Outputs

• Code or configuration changes committed according to tasks

• Test results demonstrating acceptance criteria were met

• Updated proposal/tasks capturing status and follow-ups

• execution-log.md documenting commands and evidence (template: assets/execution-log-template.md.template )

scripts/update_execution_log.py appends timestamped entries to the execution log inside openspec/changes/<change-id>/ .

Process

• Confirm scope and prerequisites via CHECKLIST.md .

• Plan work referencing affected files and dependencies.

• Implement tasks iteratively, documenting commands and results in execution-log.md via the script or manual edits.

• Run relevant tests or validation steps after each change and capture evidence in the log.

• Update artifacts and communicate completion or blockers.

Quality Gates

All checklist items must pass. If complexity grows beyond Level 1, escalate back to BMAD pathways.

Error Handling

• When prerequisites or environment setup are missing, stop and request clarity.

• If tests fail or scope expands, log findings and recommend next actions, including potential migration to BMAD development-execution.