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 restarter/lets-workflow/plugin install letsWrote 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/skills/restarter/lets-workflow/plan-workflow)<a href="https://agentmods.dev/skills/restarter/lets-workflow/plan-workflow"><img src="https://agentmods.dev/badge/skills/restarter/lets-workflow/plan-workflow/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/skills/restarter/lets-workflow/plan-workflow"><img src="https://agentmods.dev/badge/skills/restarter/lets-workflow/plan-workflow.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 | $0.00051 | $0.02313 |
| Opus 5.5 | $0.00020 | $0.00925 |
| Sonnet 5.5 | $0.00010 | $0.00463 |
| Haiku 4.5 | $0.00005 | $0.00231 |
Grade A, and why
plan-workflow scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 68 lines — stays where its author put it; the contents beside it link to each section on GitHub.
plan-workflow (PREVIEW Dynamic Workflow asset)
Preview / experimental. plan.workflow.js is the autonomous-planning chain, invoked by /lets:plan-workflow via:
Workflow({ scriptPath: "${CLAUDE_PLUGIN_ROOT}/skills/plan-workflow/plan.workflow.js", args })
It runs the autonomous design without touching the heavily-used interactive /lets:plan. Once stable (after cross-project testing), the asset folds into native /lets:plan (as a --workflow mode or equivalent) and this standalone is retired (lets-jsw00).
Why this is autonomous-only (not a re-skin of review/opinion)
/lets:review --workflow and /lets:opinion --workflow are performance levers: same result, off-context. Native /lets:plan is interactive (clarify -> explore -> approaches -> architecture -> eval -> discuss); every checkpoint forces intermediate results back into context, which kills the off-context win. So the only workflow that pays off is the autonomous whole-command form: the user front-loads a RUBRIC, judge agents make the informed picks against it, and the run returns a plan + decision log. Steer-by-rubric + approve-at-end.
Stages (off-context, NO user gate between them)
- Explore - fan out
lets:explorerper focus area (EXPLORE_SCHEMA) -> codebase map. - Approaches - one
lets:architectsynthesizes 2-4 distinct approaches from the map + goal + rubric (APPROACHES_SCHEMA). - Architect - one
lets:architectper approach -> full architecture (ARCH_SCHEMA). - Judge - a panel (default
pragmatist, backend, security- neverarchitect) scores each architecture against the rubric and picks a winner (JUDGE_SCHEMA);aggregateJudgestallies winner-votes (tiebreak: summed totals) over REAL approach ids only. - Evaluate - expert panel evaluates the winner for risks (
EVAL_SCHEMA). - Plan - one
lets:architectwrites the bite-sized plan markdown, folding in eval findings (PLAN_SCHEMA). - Plan Review - fan out
lets:architect+lets:pragmatistover the WRITTEN plan (mirror/lets:review --plan,PLAN_REVIEW_SCHEMA) -> verdict + findings. - Revise - one
lets:architectapplies the review findings -> revised plan (REVISE_SCHEMA); skipped if no findings; planMd never lost on agent error. - Plan Check - one
lets:pragmatistruns a quick 5-lens sanity pass (mirror/lets:check --plan,PLAN_CHECK_SCHEMA) -> verdict + findings; catches regressions the revise introduced. - Refine - one
lets:architectapplies the check findings -> final plan; skipped if clean.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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 Changed · +2 lines 1c8dd792ddb0
- 1mo ago First seen · 66 lines · 51 tokens per session scan A 4b88d2114985
plan-workflow is a skill published in the GitHub repository restarter/lets-workflow (17 stars, last pushed today), licensed MIT. It adds 51 tokens to every session and 2,313 once invoked, about $0.0002 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 2026-08-30.
Other skills, from other repositories
spike-consumer-adversarial
OI-3 spike harness — heavy consumer, ADVERSARIAL arm. Worst-case early-exit test: the mid-workflow Skill call has no continuation guardrail and the guidance skill ends with a final-sounding anchor. Use only when explicitly invoked by the spike harness with a TRIALID and data path.
ci-cd
A collection of practices for automating software checks and delivery. CI/CD means continuous integration and continuous delivery: changes are automatically checked, built, and potentially deployed.
experiment-review
Analyze A/B tests and experiments with statistical rigor, assess significance, perform segment analysis, and produce a clear ship/kill/extend recommendation.
run
The single lifecycle entry point — take one or more tickets and drive each through its whole lifecycle (investigate, red tests, implement, gates, review, PR, merge, archive), interleaving them. Lean by default — one worker per ticket runs red tests, implementation and the gate scripts in one worktree, then…
nazgul:heartbeat
Run one Nazgul automation-heartbeat tick — triages the work inbox and auto-starts the next objective if idle. Opt-in and default-off; fired by an optional Claude Code native scheduled agent (routine) or run by hand. Use when asked to "run a heartbeat tick", "check the inbox", or to test/debug the heartbeat.
review-work
Post-implementation gate review: run manual QA on the real surface yourself, then launch ONE gate reviewer (never a panel) to audit goal, constraints, code quality, security, missed context, and QA evidence. Use before a PR handoff or when the user explicitly asks to review completed work.