recraft

recraft is a command for coding agents from adilkalam/orca. It costs 62 tokens per session (3,316 once invoked), scanned A, original, MIT.

A coordinator for trying again after a design or interface change is judged unsuccessful. It identifies whether the problem affects one element, a whole feature, or the underlying agreement about the work, then routes it to the appropriate design workflow.

In plain words
What is it for?
Use it when a redesign, refinement, or interface implementation needs to be restarted, when a visual result is rejected, or when the user asks to start over.
Why use it?
It helps choose the right kind of redo instead of treating every design problem as a request to regenerate everything. It records user feedback for later design decisions.

Command

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 commands/adilkalam/orca/recraft
Clone the repo
git clone --depth 1 https://github.com/adilkalam/orca

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 recraft

README.md
[![agentmods](https://agentmods.dev/badge/commands/adilkalam/orca/recraft.svg)](https://agentmods.dev/commands/adilkalam/orca/recraft)
Your own site
<a href="https://agentmods.dev/commands/adilkalam/orca/recraft"><img src="https://agentmods.dev/badge/commands/adilkalam/orca/recraft.svg" alt="Measured on agentmods" height="20"></a>
Per session 62 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 3,316 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 $0.00062 $0.03316
Opus 5 $0.00031 $0.01658
Sonnet 5 $0.00012 $0.00663
Haiku 4.5 $0.00006 $0.00332

Measured 3d ago against content hash 14a45f8d85c8, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

recraft 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 3d 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.

commands/recraft.md · 158 lines

How it starts

The opening of the file, as written. The whole thing — 158 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/recraft — Try-Again Coordinator

The redo command. Thin coordinator, not generator. When work fails the eye or the user critiques it, /recraft classifies the scope of the problem and routes to the appropriate downstream flow. It does NOT generate replacement code itself.

The prior version of this command framed redo as a CI/TDD-style script loop: grep against audit rules, fail-fast, regenerate. That framing is wrong. Design has unbounded combinatorics — no rule list can enumerate every failure mode, so no script can serve as a binding gate. Taste binds through human selection at structured points (shape interview, visual-direction comp pick, /live variant accept), not through ever-larger rule lists. /recraft now reflects that.

Use /recraft when:

  • A /impeccable --craft output failed visually and the user is dictating what's wrong
  • An existing UI is being torn out and rebuilt rather than refined
  • A /refine pass produced more work than refinement and you're starting over
  • The user said any variant of "redesign / try again / this sucks / start over"

Use /refine when:

  • The current implementation is roughly right and you want a single-flag pass over it
  • The work is additive, not subtractive

Load the hub + read the contract (every invocation)

Skill("impeccable-hub") — the single home for the aesthetic (the interfaces-that-feel felt-state spine, the voice anchors, the banned rules as refusals, the preferences as positive moves, the detector contract). Loading the hub is what binds the aesthetic by construction; this file carries zero copies of those rules (#POISON_PATH re-inlining).

Then read {current-project}/.claude/PRODUCT.md (strategic) and {current-project}/.claude/DESIGN.md (visual) — both are cited in scope classification. Both files are required; if either is missing, route to /impeccable --teach or /impeccable --document before continuing.

The shared design lane (how UI-producing rebuilds run)

The UI-producing routes below (Route A → /live, Route B → re-shape + re-craft) run through the one shared lane, defined once at ~/.claude/docs/reference/design-lane.md — never re-described here (#POISON_PATH duplication; copy-pasting the lane's steps is exactly what bloated this file). The lane is: the orchestrator (this main thread) binds typed FORBIDDEN/FORWARD constraints via a cognition checkpoint and records the returned ids to {current-project}/.orca/orchestration/phase_state.json → planning.bound_constraint_ids → spawns Agent(design-builder) to produce the artifact under those ids → spawns Agent(design-validator) (fresh context) which returns GATE_VERDICT: PASS|BLOCKbranches (PASS hands back; BLOCK loops the builder with the findings, MAX N=2, then escalates to the user).

Read the full file on GitHub · 158 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. 3d ago First seen · 158 lines · 62 tokens per session scan A 14a45f8d85c8

Subscribe to this mod's changes

recraft is a command published in the GitHub repository adilkalam/orca (2 stars, last pushed 1mo ago), licensed MIT. It adds 62 tokens to every session and 3,316 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.