Agent skill

Implement

by aiblueprinthq in aiblueprinthq/ai-blueprint

Start coding or resume the approved current feature, fix, or rollback on its branch, testing each step and presenting review.

MITAuto-check passedDevOps & Cloud

Install Implement

skills CLI
$ npx skills add aiblueprinthq/ai-blueprint --skill implement -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install aiblueprinthq/ai-blueprint implement --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/aiblueprinthq/ai-blueprint.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.agents/skills/implement .claude/skills/implement && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
implement
GitHub stars
458
Token cost
~2.7k tokens
SKILL.md length
1,558 words
Files
2
Skills in repo
20
Repo updated
First seen
Licence
MIT

At a glance

Start coding or resume the approved current feature, fix, or rollback on its branch, testing each step and presenting review.

  • Works in 5 steps: Make the smallest coherent change that… → Add focused tests with logic when a test… → Run the narrowest useful check while… → …
  • Building current-feature.md
  • SKILL.md covers Start, Build loop, Verification and Final handoff
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Implement is an agent skill from aiblueprinthq/ai-blueprint. Start coding or resume the approved current feature, fix, or rollback on its branch, testing each step and presenting review. Use for /implement or building current-feature.md.

Its SKILL.md is about 2.7k tokens, which your agent loads only when the skill is triggered. The skill folder holds 2 other files (for example `reference/rollback-implementation.md`).

It sits in DevOps & Cloud. The repository describes itself as: A file-backed, spec-driven AI coding workflow framework for building real software while staying in control. The licence is MIT.

When your agent uses it

  • Building current-feature.md

Example prompts

  • “/implement”

Workflow steps

5 steps, taken from the first numbered list in SKILL.md.

  1. Make the smallest coherent change that satisfies its Done when.
  2. Add focused tests with logic when a test runner is configured. Never install
  3. Run the narrowest useful check while iterating. Do not run the full Verify
  4. Self-review the diff for contract coverage, authorization and tenant scope,
  5. Check the step box only after its code and focused check pass. Mark a repaired

What it can do on your machine

Read from SKILL.md and the folder at commit 1a0b660. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Implement loads about 2.7k tokens when it runs. Until then it costs about 47 tokens; SKILL.md has 1,558 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~47
When it runs · the whole SKILL.md, loaded when a task matches
~2.7k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from aiblueprinthq/ai-blueprint at commit 1a0b660, republished under its MIT licence (© aiblueprinthq). 1,558 words, ~2,749 tokens.

Download SKILL.mdSave it as .claude/skills/implement/SKILL.md (or your agent's skills folder). This skill also uses 1 other file; get the full folder from GitHub.
name
implement
description
Start coding or resume the approved current feature, fix, or rollback on its branch, testing each step and presenting review. Use for /implement or building current-feature.md.

implement - build the approved spec

Context reuse: Reuse any required file already loaded in project instructions or the current session. Read it again only if absent, changed, or exact current bytes or line references are needed.

The approved blueprint/context/current-feature.md is the authoritative feature packet. Implement it and stop before the work-level commit or merge.

Start

First action: Before project inspection, preflight, or any other tool call, publish the implement activity as running when blueprint/.state/ exists. Combine it with the first preflight tool batch when supported.

In one preflight batch, read:

  • blueprint/config.json
  • blueprint/context/current-feature.md
  • the current branch, status, and recent relevant log
  • the Commands section already loaded from AGENTS.md

Stop for /doctor on invalid config. Stop for /feature, /fix, or /rollback when the active spec is empty. Preserve unrelated worktree changes.

Do not reread the full project overview, coding standards, interaction guide, history, findings, or review ledger before coding. The approved spec already contains the product contract and applicable conventions. Read one targeted section only when the spec explicitly depends on a missing detail. Read findings and review state once at the final handoff.

Inspect the implementation surface in one targeted, batched read before editing. Use one additional read batch only when an exact dependency remains unknown and blocks the next change. Do not list or survey the repository, inspect unrelated examples, or run scratch environment probes. Follow the declared runtime and existing target-area patterns, then let the narrow check expose incompatibilities.

Resolve the work branch before editing. Use the exact **Branch:** value in the spec when present. For an older spec without it, combine the configured prefix for its type with the work title: lowercase ASCII letters and digits, replace each run of other characters with one hyphen, and trim edge hyphens. Feature titles come from the named build-plan item; fix and rollback titles come from their spec heading or target. Stop if the type or title is ambiguous. Create or switch to that exact branch and never implement on the default branch. On resume, start at the first unchecked build step and use git status plus the checked boxes to distinguish finished work from unfinished work.

If the spec says Type: Rollback, read and follow reference/rollback-implementation.md before changing product files. Do not load that reference for a feature or fix.

Build loop

Follow build steps in order. Build only what the spec says. If a step requires an unresolved product decision, unsafe action, missing prerequisite, or material scope expansion, stop and revise the spec instead of improvising.

Add an abstraction, dependency, service, configuration surface, compatibility layer, or security mechanism only when the approved spec or an established repository requirement needs it now. Prefer existing code, the standard library, native platform features, and installed dependencies. If the simplest complete implementation conflicts with the spec, stop and revise the spec instead of silently building a larger design.

For each step:

  1. Make the smallest coherent change that satisfies its Done when.
  2. Add focused tests with logic when a test runner is configured. Never install a runner or runtime dependency unless the spec authorizes it.
  3. Run the narrowest useful check while iterating. Do not run the full Verify command after every step when workflow.stepReview is feature; run it once after all steps. Run Verify earlier only when the step explicitly requires it or later work cannot proceed without it.
  4. Self-review the diff for contract coverage, authorization and tenant scope, error handling, accidental scope, and unrelated changes.
  5. Check the step box only after its code and focused check pass. Mark a repaired finding fixed, never closed.

With verification.logicTests: required, any logic-bearing step stops and points to /tests when no test runner is configured. Its focused logic tests must pass before the step can be checked. With verification.uiEvidence: required, a UI done-when cannot pass on build output alone. Capture the configured browser evidence, or stop and ask the user to start the required server when live evidence cannot run automatically.

With workflow.stepReview: feature, continue through passing steps and present one final review packet. With workflow.stepReview: every, stop after each step with the diff, a short explanation, evidence, and a manual try path when one exists. Continue only after approval.

Checkpoint commits are offered only when workflow.checkpointCommits is enabled and the current review gate was approved. Never commit without current approval. /complete owns the final work-level commit and merge.

Do not create a separate tool round merely to narrate a passing internal step. Keep the durable checkbox current and continue. Split a step when its diff is too large to review.

Before final verification, compare every In scope item and Done when against the finished diff. For user-facing work, inspect the reachability and error classification of each required state. Catch only known expected errors at a boundary; unexpected failures must reach the unexpected-error path. Fix any missing or contradicted contract before marking the spec verified.

Run a proportionality check before final verification: every new abstraction, dependency, service, configuration surface, compatibility layer, and security mechanism must trace to the approved spec or an established repository requirement. Remove speculative machinery this work added without weakening real trust-boundary validation, data-loss prevention, accessibility, or configured verification.

Show full SKILL.md (698 more words)Show less

Verification

After all steps pass, run the project's final automated gate once. If AGENTS.md declares a Verify command, run that exact command. Otherwise run the fallback build and tests that are actually declared. Never claim a check passed without its output.

Apply configured regular gates:

  • Audit and independent review follow qualityGates.regular.
  • Check runs for always, for behavioral work under when-behavioral, or when explicitly requested.
  • /check guide follows qualityGates.regular.tryGuide: it runs for always, for user-facing work under when-user-facing, or when explicitly requested.

Do not start a dev server. When a required runtime check needs one, ask the user to start it. Build output does not prove visual, persisted-data, authenticated, or end-to-end behavior.

If Verify or a required gate fails, repair only in-scope defects, rerun the narrow failing check, then rerun the final gate. Stop on repeated failure, missing infrastructure, or a product decision.

After final Verify and required Check pass, set the active spec to verified with every completed box checked. Then resolve independent review before the final packet:

  1. If qualityGates.regular.independentReview does not select review and no request already exists, proceed directly to the final review packet.
  2. Otherwise show the exact product/test checkpoint candidate and verified spec. Include the spec when tracked; an intentionally ignored spec uses Audit's local Spec snapshot contract without changing visibility. Obtain explicit commit approval when the exact checkpoint does not already exist, then create or use it. Configuration, including review.independentExecution: "automatic", never grants commit permission, even when normal checkpoint commits are disabled. A local-spec-only revision may reuse the same approved product HEAD after normal spec and verification gates, with a new snapshot/request and full fresh review. Do not create an empty commit for ignored spec changes.
  3. Follow /audit independent current to prepare or reuse the request and record Requested execution. For requested automatic, start and wait for the generic isolated current-runtime child instructed from the project-local Audit skill, then validate the receipt. For requested manual, or when automatic capability or access to the same local spec/snapshot is unavailable, preserve the request and stop with the manual fresh-session handoff. Treat an existing request without Requested execution as legacy manual-only: never add execution fields or run a subagent against it.
  4. Continue to the final packet only with a current passing receipt whose requested execution, actual execution, and reviewer context form an allowed pairing. Never self-review or silently skip a selected gate.

Final handoff

Read blueprint/context/findings.md and blueprint/context/review.md once. Open or fixed P0/P1 findings block /complete. Repair an open blocker as a new spec checklist step, mark it fixed after its check passes, then send it back to /audit for closure. Only the user can accept a finding.

When all steps and required gates pass:

  • Update activity to ready with /complete as the resume command.
  • Present the branch, changes grouped by area, exact checks run, how to try it, findings and independent-review state, known risks, configured gate outcomes, and /complete as the next action.

After the final packet, always offer these choices:

  1. Walk me through the implementation.
  2. Request changes.
  3. Continue to the exact next workflow command.

The final walkthrough is available with either workflow.stepReview value and regardless of workflow.checkpointCommits. It is a read-only code tour, not the manual product-review path produced by /check guide, and it is not verification.

When the user chooses the walkthrough, begin with a short map of the completed feature, then follow the spec's build steps. For each step, explain its purpose, key files and symbols, important data or control flow, and non-obvious decisions. Use file and line links when the client supports them. Do not narrate every line or reload broad project context. End by offering a focused deep dive into one named area. If the feature spans too many distinct areas for one useful pass, name the sections first and let the user choose where to begin. Remain read-only unless the user separately requests changes.

Never create an ordinary step, product, or work-level commit from this skill. The sole exception is exactly one immutable independent-review checkpoint after showing its exact candidate and receiving current explicit commit approval. Configuration never supplies that approval. Never merge, push, deploy, publish, or start unrelated work from this skill.

© aiblueprinthq, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 1 other file in .agents/skills/implement of aiblueprinthq/ai-blueprint.

  • SKILL.md
  • reference/rollback-implementation.md

Open the folder on GitHubat commit 1a0b660

Compare with similar skills

Implement next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Implement compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Implement this skillaiblueprinthq/ai-blueprint458—~2.7kAutomated safety check: PassMIT
Monitor CInrwl/nx29k5 repos~4.7kAutomated safety check: PassMIT
Terraform and OpenTofu Guideagentscope-ai/QwenPaw35k6 repos~4.2kAutomated safety check: PassApache-2.0
Vercel Optimize Auditvercel-labs/agent-skills32k8 repos~4.3kAutomated safety check: PassNone
Analyze GitHub Action Logswithastro/astro63k1 repos~1.3kAutomated safety check: PassCustom licence
Docs Learn PR Previewnetdata/netdata81k—~2kAutomated safety check: PassGPL-3.0

Similar skills

  • Monitor CI

    nrwl/nx

    Monitor Nx Cloud CI pipeline and handle self-healing fixes. An agent skill from nrwl/nx.

    29k GitHub starsUsed in 5 repos~4.7k tokens
    DevOps & CloudAuto-check passed
  • Terraform and OpenTofu Guide

    agentscope-ai/QwenPaw

    Guidance for writing and testing Terraform and OpenTofu code: module structure, naming, test approaches, CI/CD workflows, state handling and security scanning.

    35k GitHub starsUsed in 6 repos~4.2k tokens
    DevOps & CloudAuto-check passed
  • Vercel Optimize Audit

    vercel-labs/agent-skills

    Official

    Runs a metrics-first audit of a deployed Vercel project, gating investigations on real signals to produce ranked, citation-backed cost and performance recommendations.

    32k GitHub starsUsed in 8 repos~4.3k tokens
    DevOps & CloudAuto-check passed
  • Official

    Analyze recent GitHub Actions workflow runs to identify patterns, mistakes, and improvements.

    63k GitHub starsUsed in 1 repo~1.3k tokens
    DevOps & CloudAuto-check passed
  • Docs Learn PR Preview

    netdata/netdata

    Use only when the user explicitly asks to build, run, preview, inspect, or validate learn.netdata.cloud locally using the contents of a PR or documentation branch before merge.

    81k GitHub stars~2k tokensUpdated today
    DevOps & CloudAuto-check passed
  • Repo Mirror Sources

    netdata/netdata

    Inspect Netdata-org source checkouts under NETDATAREPOSDIR, or set up and synchronize that mirror when requested.

    81k GitHub stars~1.2k tokensUpdated today
    DevOps & CloudAuto-check: notes

More from aiblueprinthq/ai-blueprint

All 20 skills in this repo
  • Adopt

    aiblueprinthq/ai-blueprint

    Adopt Blueprint into an existing brownfield codebase by surveying shipped behavior and generating plans, standards, commands, adapter choices, and visibility setup.

    458 GitHub stars~2.7k tokensUpdated 2 days ago
    Auto-check passed
  • CI

    aiblueprinthq/ai-blueprint

    Set up or normalize one project Verify command and matching GitHub Actions checks while preserving existing CI, with an optional local pre-push hook.

    458 GitHub stars~2.2k tokensUpdated 2 days ago
    Auto-check passed
  • Doctor

    aiblueprinthq/ai-blueprint

    Run a Blueprint health and context check covering setup, adapters, commands, visibility, plans, overview freshness, configuration, dashboard state, and workflow drift.

    458 GitHub stars~4.5k tokensUpdated 2 days ago
    Auto-check: notes
  • Feature

    aiblueprinthq/ai-blueprint

    Turn the next, named, or numbered build-plan feature into a buildable current-feature.md spec with small steps and done-when criteria.

    458 GitHub stars~2.8k tokensUpdated 2 days ago
    Auto-check passed
  • Onboard

    aiblueprinthq/ai-blueprint

    Onboard a fresh or early scaffold after Blueprint is overlaid by tuning commands, standards, adapters, visibility, and context loading.

    458 GitHub stars~4.5k tokensUpdated 2 days ago
    Auto-check passed
  • Overview

    aiblueprinthq/ai-blueprint

    Validate and normalize project-plan.md and build-plan.md, then generate the durable project-overview.md used by agents.

    458 GitHub stars~3.8k tokensUpdated 2 days ago
    Auto-check passed

Categories

Questions about Implement

What does Implement do?

Start coding or resume the approved current feature, fix, or rollback on its branch, testing each step and presenting review. Implement is an agent skill from aiblueprinthq/ai-blueprint. Start coding or resume the approved current feature, fix, or rollback on its branch, testing each step and presenting review.

When should I use Implement?

Implement fits situations like: building current-feature.md.

How do I install Implement in Claude Code?

Run `npx skills add aiblueprinthq/ai-blueprint --skill implement -a claude-code`. Or copy the skill folder (.agents/skills/implement in aiblueprinthq/ai-blueprint) into .claude/skills/implement in your project. Claude Code loads it when a task matches its description.

How do I install Implement in Codex?

Run `npx skills add aiblueprinthq/ai-blueprint --skill implement -a codex`. Or copy the skill folder (.agents/skills/implement in aiblueprinthq/ai-blueprint) into .agents/skills/implement in your project. Codex loads it when a task matches its description.

Can I use Implement in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add aiblueprinthq/ai-blueprint --skill implement -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/implement, .gemini/skills/implement, .github/skills/implement and .opencode/skills/implement in your project.

What does Implement need to run?

SKILL.md names no scripts, command-line tools or credentials: Implement is instructions for the agent only.

Does Implement access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Implement safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Implement use?

Implement is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Implement use?

About 2.7k tokens (SKILL.md is roughly 11k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Implement?

Skills that share tags, products or a category with Implement: Monitor CI (nrwl/nx, 29k stars), Terraform and OpenTofu Guide (agentscope-ai/QwenPaw, 35k stars), Vercel Optimize Audit (vercel-labs/agent-skills, 32k stars) and Analyze GitHub Action Logs (withastro/astro, 63k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Implement?

aiblueprinthq (a GitHub organization) maintains it in aiblueprinthq/ai-blueprint, which has 458 GitHub stars. The repository holds 20 skills in this directory. The repository was last updated on October 5, 2026.

Source: aiblueprinthq/ai-blueprint on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.