Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add tzachbon/smart-ralph/plugin install ralph-specumWrote 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/commands/tzachbon/smart-ralph/design)<a href="https://agentmods.dev/commands/tzachbon/smart-ralph/design"><img src="https://agentmods.dev/badge/commands/tzachbon/smart-ralph/design/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/commands/tzachbon/smart-ralph/design"><img src="https://agentmods.dev/badge/commands/tzachbon/smart-ralph/design.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 | $0.00000 | $0.02599 |
| Opus 5.5 | $0.00000 | $0.01040 |
| Sonnet 5.5 | $0.00000 | $0.00520 |
| Haiku 4.5 | $0.00000 | $0.00260 |
Grade B, and why
design scanned grade B 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 1mo 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.
Unrestricted tool accessmediumExcessive agency
A wildcard tool grant or "run any command" leaves no least-privilege boundary at all.
allowed-tools: "*" How it starts
The opening of the file, as written. The whole thing — 193 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Design Phase
Generate technical design for the active spec after explicit requirements artifact approval. You are a coordinator, not an architect -- delegate ALL work to the architect-reviewer subagent.
Read ${CLAUDE_PLUGIN_ROOT}/references/normal-mode-gates.md before gathering context.
Checklist
Create a task for each item and complete in order:
- Gather context -- resolve spec, read requirements and research
- Interview gate -- critical frontier and approval, or authorized quick bypass
- Execute design -- dispatch architect-reviewer via team
- Artifact review -- automatic spec-reviewer loop in authorized quick mode
- Walkthrough & approval -- display summary, get user approval
- Finalize -- update state, commit, stop
Step 1: Gather Context
- If
$ARGUMENTScontains a spec name, useralph_find_spec()to resolve it; otherwise useralph_resolve_current(). Store the resolved spec directory asSPEC_PATH. - If no active spec, error: "No active spec. Run /ralph-specum:new first."
- Check the resolved spec directory exists
- Check
requirements.mdexists. If not, error: "Requirements not found. Run /ralph-specum:requirements first." - Read
.ralph-state.json, reject simultaneous exact--quickand--interactive, and normalize persistent mode withphase_gate.py mode. - When normalized
quickModeis false, require explicit requirements artifact approval before clearing its approval flag. Exact quick mode continues with the validated file. - Clear the approval flag through the locked helper while preserving every other field:
python3 "${CLAUDE_PLUGIN_ROOT}/hooks/scripts/locked-state.py" merge \ --state "$SPEC_PATH/.ralph-state.json" \ --set "awaitingApproval=false" - Read context:
requirements.md(required),research.md(if exists),.progress.md. - Run any missing applicable skill discovery pass. When research exists, pass 2 must be present.
- Run prototype record selection before design generation:
python3 "${CLAUDE_PLUGIN_ROOT}/hooks/scripts/prototype-records.py" select-downstream --base-path "$SPEC_PATH" --state "$SPEC_PATH/.ralph-state.json" --target design --target 'transition:requirements->design' --path requirements.md --path design.md
- Use only valid,
gateApproved: true, non-superseded prototype evidence that affects design. Exclude skipped, failed, inconclusive, malformed, superseded, and explicitly excluded records. - If selection reports an
activePrototypesblocker for the design transition, stop before Step 2 and report the active prototype ID, blocker reason, and resume command. - If selection reports stale requirements, research, design, or task indexes that affect this design generation, stop and route to the earliest stale phase or task. Do not generate design from stale artifacts.
- Read the matching
targetDecisionsentries. Proven unrelated work may continue only when each entry hasproofAvailable: trueandeligible: true; this proves no active blocker, stale input, or approved transfer path overlaps design. Missing or unavailable proof blocks conservatively.
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.
- 1mo ago First seen · 193 lines · 0 tokens per session scan B ffb0dd2c1f64
design is a command published in the GitHub repository tzachbon/smart-ralph (559 stars, last pushed 1mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 2,599 tokens. A static security scan graded it B with 1 finding (unrestricted tool access). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.
Other commands, from other repositories
resume-project
Run resume-project now — Now / Blocked / Next. Not a cheat sheet.
backlog
PitWay: Capture out-of-scope work discovered mid-task, without expanding current scope.
frankx-ai-build
Comprehensive FrankX.AI website build session - plan, create, review, deploy.
publish
Publish a reviewed content packet through the canonical publisher and verify the remote result.
acos-swarm
Initialize a coordinated swarm of agents for complex, multi-file tasks.
plan-update
Update an initiative's status, progress, tasks, or add agent commentary.