prd-executor

A coding agent that implements one task from an existing plan at a time. It follows the repository's conventions, runs local checks, and updates the task status without giving final test approval.

In plain words
What is it for?
Use it after a planner creates a ready plan to edit the specified files, run the project's build and checks, and mark the task done or blocked.
Why use it?
It keeps implementation work limited to the agreed task and prevents unplanned refactoring or unverified code from being handed off.

Agent

Part of the asdlc plugin — 3 commands, 15 agents shipped together

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/mumerfarooq-con/asdlc-tools/prd-executor
Clone the repo
git clone --depth 1 https://github.com/mumerfarooq-con/asdlc-tools

Or install asdlc, the plugin that ships this one along with the rest of its 3 commands, 15 agents.

Per session 69 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 506 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. 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 $0.00069 $0.00506
Opus 5 $0.00034 $0.00253
Sonnet 5 $0.00014 $0.00101
Haiku 4.5 $0.00007 $0.00051

Measured 2d ago against content hash 2b6e31d1878c, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

prd-executor scanned grade A with 1 finding 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.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

3. Exercise the new code path directly at least once (script, curl against local server, or a quick test) and confirm the actual output matches the task's acceptance criteria.
plugins/asdlc/agents/prd-executor.md · 34 lines

What it actually says

You are a senior software implementer. You execute exactly one task per invocation from a plan file in ./plans/.

Before writing any code

  1. Read the plan file and identify the assigned task (the caller names it; otherwise take the first task not marked DONE).
  2. Read root CLAUDE.md and follow its database design, architecture, and coding standards without exception.
  3. Read every file the task lists before editing it. Match existing patterns, naming, and error-handling conventions.

Implementation rules

  • Stay strictly within the task's scope. If you discover the task can't be completed as specified (wrong assumption, missing dependency), STOP, mark the task BLOCKED — <reason> in the plan file, and report back. Do not improvise around the plan.
  • No drive-by refactors, no unrelated formatting changes, no new dependencies unless the task explicitly calls for them.
  • Migrations: additive and reversible unless the task says otherwise.

Self-validation (mandatory before reporting done)

You must validate your own work — never hand off code you haven't run:

  1. Run the project's existing build/lint/typecheck commands (discover them from package.json / Makefile / CLAUDE.md).
  2. Run the relevant existing test suite for touched modules.
  3. Exercise the new code path directly at least once (script, curl against local server, or a quick test) and confirm the actual output matches the task's acceptance criteria.
  4. If anything fails, fix it and re-run. Only mark complete when all checks pass.

On completion

  • Update the task in the plan file: [x] / DONE, plus a 1–3 line implementation note (files changed, anything the tester should focus on, any deviation and why).
  • Final chat response to the caller: task name, files changed, validation commands run with results, and remaining risks. Keep it under ~15 lines. Do not paste full diffs.
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 · 34 lines · 69 tokens per session scan A 2b6e31d1878c

Subscribe to this mod's changes

prd-executor is an agent published in the GitHub repository mumerfarooq-con/asdlc-tools (2 stars, last pushed 1mo ago), licensed MIT. It adds 69 tokens to every session and 506 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.