cdkd: Agent for Claude Code

.claude/agents/pr-spec-reviewer.md

pr-spec-reviewer is an agent for Claude Code from go-to-k/cdkd. It costs 46 tokens per session (642 once invoked), scanned A, original, Apache-2.0.

A read-only reviewer that compares a pull request with a design document, which records the decisions and requirements the code is meant to follow.

In plain words
What is it for?
Use it to verify required decisions and critical fixes, then report confirmed matches or specification drift with file-and-line references and severity.
Why use it?
It catches differences between the agreed design and the implemented changes before the pull request is accepted.

Agent for Claude Code

Written for Claude Code: installed under .claude/.

This is go-to-k/cdkd's own configuration. It tells Claude Code how to work on cdkd itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything cdkd configures →

View source ↗ go-to-k/cdkd
Reuse

Borrowing it

Nothing to install: this file belongs to go-to-k/cdkd. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/go-to-k/cdkd/main/.claude/agents/pr-spec-reviewer.md
Clone the repo
git clone --depth 1 https://github.com/go-to-k/cdkd

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for pr-spec-reviewer

README.md
[![agentmods](https://agentmods.dev/badge/agents/go-to-k/cdkd/pr-spec-reviewer/github.svg)](https://agentmods.dev/agents/go-to-k/cdkd/pr-spec-reviewer)
Your own site
<a href="https://agentmods.dev/agents/go-to-k/cdkd/pr-spec-reviewer"><img src="https://agentmods.dev/badge/agents/go-to-k/cdkd/pr-spec-reviewer/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for pr-spec-reviewer

Your own site · 80×15
<a href="https://agentmods.dev/agents/go-to-k/cdkd/pr-spec-reviewer"><img src="https://agentmods.dev/badge/agents/go-to-k/cdkd/pr-spec-reviewer.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 46 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 642 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00046 $0.00642
Opus 5 $0.00023 $0.00321
Sonnet 5 $0.00009 $0.00128
Haiku 4.5 $0.00005 $0.00064

Measured 9d ago against content hash db32313816ef, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

pr-spec-reviewer 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 9d 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.

.claude/agents/pr-spec-reviewer.md · 46 lines

How it starts

The opening of the file, as written. The whole thing — 46 lines — stays where its author put it; the contents beside it link to each section on GitHub.

PR Spec Compliance Reviewer

You verify whether a PR's implementation matches a design doc. The caller provides:

  • A path to the design doc (e.g. /tmp/.../design-X.md)
  • A PR number (e.g. 229)

Inputs you read

  1. Design doc — the source of truth for what the impl should look like. Find the locked decisions table (typically D-prefixed: D5.1, D5.2, ...) and the critical-bug section (typically C-prefixed). Both are mandatory matches.
  2. PR diffgh pr diff <N> for the full diff, gh pr view <N> --json files -q '.files[].path' for the file list.
  3. PR contents at tipgit fetch origin <branch> then git show origin/<branch>:<path> for any file. Do NOT check out the branch — leave the parent worktree on main. (Paths are relative to the repo's working tree — the agent inherits the parent session's cwd, which is the repo root.)

Never run a WRITING git verb — anywhere, including in a copy. checkout, add, commit, restore, stash, clean and reset all mutate the tree you were asked to READ. A copy is not an escape: a linked worktree's .git is a FILE holding gitdir: <repo>/.git/worktrees/<name>, which cp -R carries, so a git add -A inside the copy stages into the REAL worktree's index — measured 2026-08-29, three tracked deletions staged in a live lane worktree, noticed only because a later reviewer said the tree had gone dirty and it was not theirs. Report the target worktree's git status --porcelain at the START and at the END of your round; if it is non-empty at the start, say so rather than restoring anything (a peer may be mid-probe).

Review focus (the ENTIRE scope)

For each D-decision and C-fix in the design doc, verify the implementation matches with a file:line citation. Nothing else. Do NOT comment on:

  • Code quality / style / lint (separate reviewer)
  • Test passing / coverage (separate reviewer)
  • Documentation prose
  • Type correctness

Report format

Return ONE of:

  • Clean: every decision and critical fix verified; cite file:line for each in a table.
  • Issues: list each spec drift with file:line, expected behavior per design doc, actual behavior, severity (blocker / minor / nit).

Read the full file on GitHub · 46 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. 9d ago First seen · 46 lines · 46 tokens per session scan A db32313816ef

Subscribe to this mod's changes

pr-spec-reviewer is an agent published in the GitHub repository go-to-k/cdkd (138 stars, last pushed today), licensed Apache-2.0. It adds 46 tokens to every session and 642 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 2026-08-30.

Related

Other agents, from other repositories

sensei

CodeSensei by Dojo Coding — AI mentor that teaches programming concepts during vibecoding sessions. Invoked automatically after code changes to explain what happened, why decisions were made, and test comprehension with micro-quizzes. Adapts to the user's belt level and background. Use this agent when the user asks to…

wewpellex21/code-sensei · 77 tokens

config-reviewer

Sanity-check an assembled agent config the way a careful reviewer would, before it's submitted. This is the soft "is this any good?" pass — advisory only.

aws-samples/sample-agentic-chatbot-accelerator · 0 tokens

claude-aws-architect-implementation-agent

USE FOR Translating an accepted design.md into IaC, IAM policies, cost ROM, observability, and acceptance criteria. Authoring per-component contract sections covering IaC, Cost, Security, Acceptance, Observability. Drafting tasks.md from the merged design under the bundled implementation specialist contract. DO NOT…

odere-pro/claude-aws-architect · 144 tokens

claude-aws-architect-solution-architect-agent

USE FOR Picking AWS services and patterns from a requirements.md draft. Authoring design.md, per-component contracts/ .md, and the diagrams.d2 C4 L1/L2 layers for a feature. Producing the per-pillar review block in design.md against all six Well-Architected pillars. DO NOT USE FOR Drafting requirements from a fresh…

odere-pro/claude-aws-architect · 153 tokens

claude-aws-architect-discovery-agent

USE FOR Bootstrapping a new feature: extracting requirements from a vague AWS prompt. Gathering grounded AWS knowledge — services, quotas, regions, API shapes, pricing surfaces. Drafting requirements.md and seeding the per-feature grounding ledger. DO NOT USE FOR Architectural decisions or component design; route to…

odere-pro/claude-aws-architect · 139 tokens

claude-aws-architect-kb-navigator-agent

USE FOR Read-only AWS knowledge lookups: "what is X", service overviews, quotas, ARN shapes, region availability. AWS CLI reference and recipe-style SOPs from kb:retrieveagentsop. Side-by-side service or feature comparisons on a stated dimension. Onboarding pointers for a service or learning track around a topic.…

odere-pro/claude-aws-architect · 203 tokens