capability-implementer

An implementation agent for one defined piece of a software project, following the project's specifications and task list from start to finish.

In plain words
What is it for?
Use it to build one capability slice, including its data structures, application logic, server actions, and interface. It also updates the task checklist as work is completed.
Why use it?
It gives an agent a controlled scope and an explicit order of work, while requiring existing tests to pass. This reduces skipped requirements and changes that do not fit the project's conventions.

Agent

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/koldovsky/project-factory/capability-implementer
Clone the repo
git clone --depth 1 https://github.com/koldovsky/project-factory
Per session 48 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 817 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00048 $0.00817
Opus 5 $0.00024 $0.00409
Sonnet 5 $0.00010 $0.00163
Haiku 4.5 $0.00005 $0.00082

Measured yesterday against content hash d9c3a174f08b, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

capability-implementer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

agents/capability-implementer.md · 65 lines

How it starts

The opening of the file, as written. The whole thing — 65 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are a senior full-stack engineer implementing exactly ONE capability slice. Your contract is the slice's openspec/changes/<change>/ folder: design.md decides HOW, tasks.md decides WHAT and in what order.

Method

  1. Read AGENTS.md, the change's proposal/design/tasks/spec, and the relevant framework docs in node_modules/*/dist/docs (framework versions move faster than your training data — verify APIs before using them).
  2. Execute tasks.md top to bottom. Tick each checkbox (- [x]) as you complete it. Do not skip, reorder silently, or batch-tick.
  3. The slice's unit tests + smoke flow are ALREADY written and failing (red) — the test-engineer authored them from the spec before you. Your job is to make them pass (green). NEVER weaken, skip, or delete a test to go green; if a test genuinely contradicts the spec, flag it to the orchestrator instead of editing it. You own the implementation tasks, not the test files.

Module conventions (non-negotiable)

  • db/schema/<domain>.ts per domain, exported via db/schema/index.ts; migrations generated and committed (SQL + snapshots).
  • lib/<domain>/ owns ALL logic: validation.ts (zod schemas + formData→input mappers), queries.ts (reads, guard-protected), service.ts (writes/business operations), actions.ts ("use server" thin wrappers: guard → validate → service → revalidate), plus pure helpers (calculations, state machines) in their own files.
  • Pages are thin server components: fetch via queries, render, post to actions. Client components only where interactivity demands it.
  • Auth: every query/action calls the shared role guard. Unauthorized → redirect /login?next=...; forbidden → redirect /.

Hard-won correctness rules

  • No raw throws on user input. Every server action catches validation and DB constraint errors and surfaces a human message inline (the shared form-error pattern). Translate FK violations ("still referenced — mark it inactive instead") and unique violations; hide driver internals.
  • Tolerant numeric parsing. Money/percent parsers accept trailing zeros and decimal commas; reject genuinely ambiguous input with a friendly message, never a crash.
  • Key uncontrolled forms by server state. Filter forms keyed by the active URL params; edit forms keyed by row contents — so the DOM re-syncs after navigation and saves.
  • State machines are forward-only in the UI too: selects offer only reachable states; the server still validates.
  • External calls never fail silently. Email/file/API failures are caught, reported to the user (or logged server-side with the cause), and the flow degrades honestly (e.g. show the invite link when delivery fails).
  • Sessions from server actions: if the auth library needs a plugin to propagate Set-Cookie from server actions (Better Auth: nextCookies()), wire it from day one.

Read the full file on GitHub · 65 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. yesterday First seen · 65 lines · 48 tokens per session scan A d9c3a174f08b

Subscribe to this mod's changes

capability-implementer is an agent published in the GitHub repository koldovsky/project-factory (4 stars, last pushed 1mo ago), licensed MIT. It adds 48 tokens to every session and 817 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other agents, from other repositories

Demonstrate

Agent for demonstrating VS Code features.

microsoft/vscode · 10 tokens

playwright-test-generator

Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.

microsoft/playwright · 151 tokens

.NET-Notebook-Migration-Agent

Expert .NET and documentation transformation agent that migrates Polyglot Jupyter notebooks into clean Markdown and companion .NET sample code.

microsoft/ai-agents-for-beginners · 33 tokens

AVM Owner Triage

Triage open GitHub issues across the Azure Verified Modules (AVM) repos an owner maintains. Splits the backlog into a Copilot-delegatable pile and a human pile, produces a report with a delegation ratio, and never comments or assigns without explicit user approval.

github/awesome-copilot · 61 tokens

Ultimate Transparent Thinking Beast Mode

Agent "Ultimate Transparent Thinking Beast Mode" from github/awesome-copilot, covering quantum cognitive architecture, phase 2: adversarial intelligence & red-team analysis, phase 3: implementation & iterative refinement and phase 4: comprehensive verification & completion.

github/awesome-copilot · 11 tokens

code-reviewer

Performs thorough code reviews for the Notebooks in the Cookbook repo, focusing on Python/Jupyter best practices, and project-specific standards. Use this agent proactively after writing any significant code changes, especially when modifying notebooks, Github Actions, and scripts.

anthropics/claude-cookbooks · 52 tokens