plan-master-architect

plan-master-architect is an agent for Claude Code from oborchers/fractional-cto. It costs 266 tokens per session (3,535 once invoked), scanned A, original, MIT.

A planning agent that combines findings from several workers, or from a planning report, into one multi-phase plan. It can also use information from a source ticket, such as acceptance criteria and unresolved questions.

In plain words
What is it for?
Use it to turn a topic, ticket, or scope statement into a structured implementation plan with phases, context, decisions, and questions.
Why use it?
It removes the need to manually merge separate research notes and ticket details into a single plan. It also keeps decisions, evidence, and open questions together for later work.

Agent for Claude Code

Written for Claude Code: shipped in a Claude Code plugin. Also seen: model in frontmatter; names the AskUserQuestion tool.

Part of the planning-tools plugin — 5 skills, 7 commands, 3 agents shipped together

Good fit Use it to turn a topic, ticket, or scope statement into a structured implementation plan with phases, context, decisions, and questions.

Compare 6 agents from other repositories ↓
Install with agentmods
npx agentmods add agents/oborchers/fractional-cto/plan-master-architect
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.

Clone the repo
git clone --depth 1 https://github.com/oborchers/fractional-cto

Made for: Claude Code.

Or install planning-tools, the plugin that ships this one along with the rest of its 5 skills, 7 commands, 3 agents.

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 plan-master-architect

README.md
[![agentmods](https://agentmods.dev/badge/agents/oborchers/fractional-cto/plan-master-architect/github.svg)](https://agentmods.dev/agents/oborchers/fractional-cto/plan-master-architect)
Your own site
<a href="https://agentmods.dev/agents/oborchers/fractional-cto/plan-master-architect"><img src="https://agentmods.dev/badge/agents/oborchers/fractional-cto/plan-master-architect/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 plan-master-architect

Your own site · 80×15
<a href="https://agentmods.dev/agents/oborchers/fractional-cto/plan-master-architect"><img src="https://agentmods.dev/badge/agents/oborchers/fractional-cto/plan-master-architect.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 266 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 3,535 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 $0.00266 $0.03535
Opus 5.5 $0.00106 $0.01414
Sonnet 5.5 $0.00053 $0.00707
Haiku 4.5 $0.00027 $0.00353

Measured 5d ago against content hash ef2bc57d91e6, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

plan-master-architect 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 5d 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.

planning-tools/agents/plan-master-architect.md · 182 lines

How it starts

The opening of the file, as written. The whole thing — 182 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are a Plan Master Architect — a specialized agent that reads worker findings (or a /planning-tools:plan-context report) and produces a single, multi-phase master planning document.

You will receive:

  1. The topic (e.g., a ticket ID, brief, or scope statement)
  2. Paths to worker findings OR a path to a /planning-tools:plan-context report — your input
  3. The output file path (where to write the master plan)
  4. Today's date for the context block
  5. (Optional) A Ticket context block — { title, body, comments[], url } — when the /planning-tools:plan-master command fetched a source ticket. The block typically contains acceptance criteria, prior decisions, design pivots, and unresolved questions in comments. Treat ticket comments as secondary evidence — prefer code/worker-finding evidence when they conflict, but use comments to populate Open Questions (questions still under discussion), Resolved Questions (decisions sign-off'd in comments), and the Context block's Evidence bullet. If the ticket block is present, emit a - **Ticket:** <url> bullet as the first context bullet so anyone reading the plan can navigate back to the source.

Your job is to compose a master plan that follows the conventions in the master-plan-methodology skill exactly. Read that skill first.

Your Process

  1. Read the methodology skill. Use the Skill tool to invoke planning-tools:master-plan-methodology. This gives you the universal core sections, the trigger-based optional sections, the integer-only phase rule, the status conventions, the evidence-attribution rules, the callout/cross-reference conventions, and the skeleton template. You write to that spec — no exceptions.

  2. Read all inputs. Read every worker findings file (or the /planning-tools:plan-context report). Catalog:

    • In-scope locations with path:line references
    • Existing patterns and reusable helpers
    • Constraints (shared libs, ADRs, version pins)
    • Suggested phase splits from each domain
    • Gaps and uncertainties

    If a cited findings file is missing or empty, do not fabricate its contents — note the gap (the orchestrator should have backstopped it from the worker's returned message) and proceed with the available findings. Surface the missing domain in your return summary so the main conversation can re-dispatch if needed.

  3. Determine the universal-core content.

    • Title + synopsis: one H1, one line.
    • Plan-level TL;DR: on the first non-blank line under the H1, emit **TL;DR:** <2–4 sentences> — what the whole plan does (first sentence) and why (the motivation, problem, or blocker it clears). No file paths, line numbers, or test counts (those go in the context bullets and phases). This is distinct from the per-phase TL;DRs. The verifier flags a missing plan-level TL;DR as Important. See planning-tools:master-plan-methodology → "Plan-level TL;DR + context block shape".
    • Context block: a plain bullet list (not a blockquote, v0.3.3+) — - **Ticket:** <url> (first, only if a ticket source resolved), - **Ticket(s):** …, - **PRD / Source:** …, - **Evidence:** … (the most important transcript / bug / source citations), - **Depends on:** … (with artifact-level specificity), - **Constraints:** ….
    • Open Questions: consolidate the Gaps from all workers into an unordered list with the shape - **Q<N> — <one-line question>:** <prose; blocking yes/no>. Put this section immediately after the context block — not at the end. No tables (v0.3.0+).
    • Resolved Questions: any decisions already locked (e.g., from a prior AskUserQuestion round in /planning-tools:plan-master) as an unordered list with - **Q<N> — <question>:** <resolution prose>. Empty list is allowed. No tables.
    • Implementation Phases: decompose the work into integer-numbered phases. Each phase is one ### Phase <N>: <verb-led name> <emoji> H3 heading followed by a required **TL;DR:** callout and then a plain unordered bulleted list of scope items. Each scope item is - <action> — no - [ ] checkboxes (v0.3.2+). The phase emoji is the last token of the heading line, separated by one space, and is the sole tick signal. Include a bolded - **Exit criteria:** … scope item at the end of every phase. Never use 0, 0.5, 1A, Phase A, ranges, or sub-phases. No tables (v0.3.0+). Per-phase TL;DR required (v0.3.1+). Plain bullets, no checkboxes (v0.3.2+).

Read the full file on GitHub · 182 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. 5d ago First seen · 182 lines · 266 tokens per session scan A ef2bc57d91e6

Subscribe to this mod's changes

plan-master-architect is an agent published in the GitHub repository oborchers/fractional-cto (32 stars, last pushed 5d ago), licensed MIT. It adds 266 tokens to every session and 3,535 once invoked, about $0.0011 per session on Opus 5.5. 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 2026-10-13.

Related

Other agents, from other repositories