apex · melvynx — Claude Code skill: staged implementation with checks
Runs a task through Analyze → Plan → Execute → eXamine with checkpoints, scoped delegation and evidence. Fits features, bug fixes and migrations.
npx skills add Melvynx/aiblueprint --skill apex -a claude-codeWhat apex does
The skill lays out a fixed pipeline of nine stages: contract, analysis, plan, execution, validation, examination, tests, runtime proof, handoff. Each one lives in its own file under steps/, and the agent reads only the current step instead of pulling the whole process into context. Run state goes through scripts/apex-state.py into .agents/apex/runs/<run-id>/; secrets do not belong there.
Flags tune the run. -a/-A pick low or standard interaction, -t/-T switch new tests on or off, -v/-V control runtime proof, -e/-E set the budget, -pr/-PR decide on a pull request. Resume an interrupted run with -r <id>. Flags carry intent; they do not dictate a specific implementation.
Only bounded, independent chunks may be delegated, and whatever a subagent returns stays untrusted until the coordinator reads the diff. The run closes when the implementation exists, acceptance criteria have fresh evidence, regressions introduced along the way are fixed, and delivery actions were read back from the system.
What it helps with
- carry a feature from analysis to a verified result
- split a migration into checkpointed stages
- attach tests and runtime proof according to risk
- hand a bounded slice of work to a subagent
- resume a stopped run from a saved checkpoint
Example prompts for the agent
- Run apex on this task: add pagination to the orders list
- Start apex with
-tand-vso tests and runtime proof are required - Continue the apex run:
-r <id>
How to install apex
1Skills CLI (Vercel)
npx skills add Melvynx/aiblueprint --skill apex -a claude-code2Manually: put the files in place
~/.claude/skills/apex/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/apex/
1Skills CLI (Vercel)
npx skills add Melvynx/aiblueprint --skill apex -a codex2Manually: put the files in place
.agents/skills/apex/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add Melvynx/aiblueprint --skill apex -a cursor2Manually: put the files in place
.cursor/skills/apex/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install Melvynx/aiblueprint apex2Manually: put the files in place
.github/skills/apex/SKILL.md1Skills CLI (Vercel)
npx skills add Melvynx/aiblueprint --skill apex -a gemini-cli2Platform CLI
gemini skills install https://github.com/Melvynx/aiblueprint --path agents-config/skills/apexManually: put the files in place
.devin/skills/apex/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add Melvynx/aiblueprint --skill apex -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about apex
How do I install apex in Claude Code?
Run npx skills add Melvynx/aiblueprint --skill apex -a claude-code. The skill lands at ~/.claude/skills/apex/SKILL.md.
What does apex do?
It drives a code change through Analyze → Plan → Execute → eXamine, saves run state, and will not finish until acceptance criteria are backed by current evidence.
Where does apex keep run state?
Under .agents/apex/runs/<run-id>/, written by scripts/apex-state.py. Secrets must not go into that state.
Similar skills
All skills for Agent setup: memory, context, orchestration →apex by other authors comes first.
