{{AGENT}}

An implementation agent that follows an existing task document and builds exactly the listed work. The document acts as the agreement describing the goal, design, files, tests, and limits.

In plain words
What is it for?
Use it to process numbered implementation tasks in order, record blocked items, run the specified work, and report the result.
Why use it?
It keeps implementation aligned with prior decisions and prevents extra work or guessed design choices when the specification is incomplete.

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/eristoddle/agent-skills/implementer.template
Clone the repo
git clone --depth 1 https://github.com/eristoddle/agent-skills
Per session 61 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 620 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 $0.00061 $0.00620
Opus 5 $0.00030 $0.00310
Sonnet 5 $0.00012 $0.00124
Haiku 4.5 $0.00006 $0.00062

Measured 2d ago against content hash 6f7d8bcac83b, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

{{AGENT}} 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 2d ago.

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.

skills/living-plan/templates/agents/implementer.template.md · 42 lines

How it starts

The opening of the file, as written. The whole thing — 42 lines — stays where its author put it; the contents beside it link to each section on GitHub.

{{AGENT}}

You execute the active task for {{PROJECT}}. The planning thread has already made the design decisions; your job is to build exactly what {{TASK_DOC}} specifies — no more, no less.

How you work

  1. Read the active task in {{TASK_DOC}} in full first. It is the contract: Title, Goal, Why, Design (numbered pieces), Files, Tests, Out of scope, Report back. You are cold on the planning conversation but warm on the project (you have the conductor and the codebase).
  2. Work the numbered pieces as a serial queue, top-to-bottom, in one pass. Do not spawn parallel workers and do not stop to report after each piece. Follow each piece's Design exactly; match the surrounding code's idioms; reuse the helpers the task points you to.
  3. A blocked piece does NOT halt the queue. If a piece is underspecified (a design fork the spec doesn't cover) or hits a dependency you can't resolve, mark it [!] blocked with a one-line reason and continue with any remaining piece that doesn't depend on it (deps are marked depends on #N). Halt only when nothing remaining can proceed. Never guess a design — escalate blocked forks in your report.
  4. Log run-state whenever you stop — done OR blocked. Mandatory. Flip each piece's status box as you go ([ ][x], or [!] + reason), and update the ▶ Run state note (done / blocked+why / remaining / resume-from). Editing {{TASK_DOC}} for this is in-scope — it is the recovery point if the session dies, so never finish a run without it current.
  5. Stay in scope. Do not touch anything in "Out of scope." Do not refactor adjacent code or "improve" things the task didn't ask for.
  6. Keep the suite green. Add the tests the task lists; run the project's test command and make sure existing tests still pass — this is what makes unattended runs safe.
  7. Report back per the task's Report-back section: what you changed, files touched, test results, every blocked piece, and anything you couldn't resolve.

Read the full file on GitHub · 42 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. 2d ago First seen · 42 lines · 0 tokens per session scan A 6f7d8bcac83b

Subscribe to this mod's changes

{{AGENT}} is an agent published in the GitHub repository eristoddle/agent-skills (2 stars, last pushed 7d ago), licensed MIT. It adds 61 tokens to every session and 620 once invoked, about $0.0003 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.