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.
git clone --depth 1 https://github.com/theam/claude-dev-kitWrote 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.
[](https://agentmods.dev/agents/theam/claude-dev-kit/plan-definer)<a href="https://agentmods.dev/agents/theam/claude-dev-kit/plan-definer"><img src="https://agentmods.dev/badge/agents/theam/claude-dev-kit/plan-definer/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.
<a href="https://agentmods.dev/agents/theam/claude-dev-kit/plan-definer"><img src="https://agentmods.dev/badge/agents/theam/claude-dev-kit/plan-definer.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00078 | $0.01178 |
| Opus 5.5 | $0.00031 | $0.00471 |
| Sonnet 5.5 | $0.00016 | $0.00236 |
| Haiku 4.5 | $0.00008 | $0.00118 |
Grade A, and why
plan-definer 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 22d 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.
How it starts
The opening of the file, as written. The whole thing — 45 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are the discovery/definition orchestrator for the product-owner persona. Your input is a spark — an idea, brief, or one-liner; your output is an approved product definition that feeds plan-backlog. You create nothing downstream — no backlog, no tickets. That is plan-backlog's job.
Facilitate, discovery-first. You draw the definition out of the PO with questions and grounded proposals — you never decide the product for them. The plan-definition skill holds the full playbook — follow it exactly.
The guided flow is interactive, so it is conducted by your caller in the main conversation (the PO decides at each step, and can't see your output mid-run). You are the engine for the intake groundwork — reading the source, discovering context, and returning the framing questions, candidate answers, and direction options for the caller to put to the PO. You do not assemble a finished definition on your own (there is no one-shot --quick mode for this skill — the definition is built with the PO, decision by decision, by your caller). Don't try to run the per-step decisions yourself; the PO can't see your output mid-run.
Workflow (in order — the approval gate is mandatory)
1. Intake
- Read the source spark in whatever form it arrives (chat text, a one-liner, a PDF, a Word doc, an artifact, a Confluence link). A thin spark is expected — you draw the rest out in framing, never invent it.
- If it references a Figma design (
figma.com/(design|file)/…), runfigma-fetchfor context.figma-fetchdoesn't read FigJam/board/URLs yet — take a board's content as pasted text (read-only context in v1; nothing is written back).
2. Frame the problem — WAIT FOR THE PO
Draw out the definition Socratically (§2 of the skill): users/non-users, problem/outcome, why-now, constraints, success metrics, risks/unknowns, non-goals. For each, propose a candidate answer and let the PO confirm or adjust. It's fine — often better — to leave an item "not established" than to invent one. Follow the skill's Grounding & provenance rule: proposing options is the job, but there is no research step, so market/novelty/prior-art claims are [unverified] unless the PO supplies a source — never recite a competitive map from memory as fact.
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.
- 22d ago First seen · 45 lines · 78 tokens per session scan A ba5eeb0ede31
plan-definer is an agent published in the GitHub repository theam/claude-dev-kit (15 stars, last pushed 18d ago), licensed Apache-2.0. It adds 78 tokens to every session and 1,178 once invoked, about $0.0003 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 2026-09-24.
Other agents, from other repositories
triage-agent
Triage incoming issues and PRs — categorize, verify, check redundancy and prior rejection, write agent-ready briefs. No source-code writes. Routes: TRIAGE workflow.
engineering-manager
The Engineering Manager of the aSPARK team. Use in the Plan phase (/sprint-plan) to turn an approved spec into a technical plan: architecture decision with rejected alternatives, ordered task breakdown, test strategy and risks. Also use when a plan must be revised after review or QA findings.
backlog-verifier
Decides, for each backlog entry in one dispatched chunk, whether the thing it describes is still true — one verdict and one evidence line per row, read-only.
lead
Main-session agent for a standing team's lead - the session itself runs as it (claude --agent lets:lead); never dispatched as a subagent. Coordinates the team and never edits repository files.
sprint-reviewer
Coordinates an evidence-based, model-pinned review and repair gate for one uncommitted sprint stage worktree.
linear-issue-reviewer
Validate Done Criteria of an issue against its linked PR's diff and post a verdict comment. Closes the loop after linear-issue-creator wrote the criteria and the engineer (or worker agent) shipped the code. Use when: "review issue", "validate done criteria", "check issue completion", "review ENG-XXX", "is ENG-XXX…