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.
npx agentmods add commands/harshvardhan86/claude-wave-plugin/wave-startgit clone --depth 1 https://github.com/Harshvardhan86/claude-wave-pluginWrote 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/harshvardhan86/claude-wave-plugin/wave-start)<a href="https://agentmods.dev/commands/harshvardhan86/claude-wave-plugin/wave-start"><img src="https://agentmods.dev/badge/commands/harshvardhan86/claude-wave-plugin/wave-start.svg" alt="Measured on agentmods" 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 | $0.00075 | $0.00763 |
| Opus 5 | $0.00037 | $0.00381 |
| Sonnet 5 | $0.00015 | $0.00153 |
| Haiku 4.5 | $0.00007 | $0.00076 |
Grade A, and why
wave-start scanned grade A 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 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
allowed-tools: "Bash(git:*), Bash(npm:*), Bash(pnpm:*), Bash(yarn:*), Bash(npx:*), Bash(curl:*)" How it starts
The opening of the file, as written. The whole thing — 60 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Start a new wave
Context
- Working dir: !
pwd - Git repo: !
git rev-parse --is-inside-work-tree 2>/dev/null && echo "Yes" || echo "No" - Branch: !
git branch --show-current 2>/dev/null || echo "n/a" - Dirty tree: !
git status --porcelain 2>/dev/null | head -1 | grep -q . && echo "Yes (commit or stash before starting a wave)" || echo "Clean"
Instructions
The user invoked /wave-start with arguments: $ARGUMENTS
Step 1 — parse the mode
- If
$ARGUMENTSstarts with--demo, set mode =demo. The feature description is everything after--demo. - Otherwise, set mode =
full. The feature description is the entire$ARGUMENTS.
If the feature description is empty, refuse and remind the user of the syntax:
/wave-start "<feature description>" # full v2 framework (17 base phases + conditional [DR]/[CR] gates)
/wave-start --demo "<feature description>" # trimmed 5-phase demo subset
Step 2 — invoke the wave-orchestrator skill
Invoke the wave-orchestrator skill, passing:
- The mode (
fullordemo) - The feature description
The orchestrator owns everything from here. Do not implement, dispatch, or instruct further yourself. The orchestrator's SKILL.md defines exactly how each mode runs.
Step 3 — observe the rules
Both modes enforce:
- Lean, dedicated context per dispatch (no combined phases)
- Strict TDD: RED before GREEN, watch tests fail before any code is written
- All bugs P0; user decides how to fix
- "tsc passes" is NEVER hand-off evidence — live stack + screenshot is
- Visual assertions use
getComputedStyle()andboundingBox, nevertoBeVisible()alone - No AI traces in files, commits, or PRs
full mode additionally:
- Loads
framework/SKILL.mdand selectedframework/references/*.mdper phase - Routes dispatches to Opus / Sonnet / Haiku per
framework/references/10-model-routing.md - Uses
isolation: "worktree"on writing dispatches - Runs Dashboard and long Playwright via
run_in_background - Engages
[DR]at Phase 1.5 whenever the wave ships UI; skips it silently otherwise - Includes BC, SEA, BSEA, DS, OA, TEET-TC, BTEET, BTEET-X, version bump, commit, cleanup, checkpoint, dashboard
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.
- 5d ago First seen · 60 lines · 75 tokens per session scan A 948e771d6480
wave-start is a command published in the GitHub repository Harshvardhan86/claude-wave-plugin (8 stars, last pushed 3mo ago), licensed MIT. It adds 75 tokens to every session and 763 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
resume
Resume an interrupted workflow from where it left off.
release
Release the plugin — bump version across 4 files, then optionally commit, tag, push, and create a GitHub Release.
fierce-option-panel
Default-on interview option-quality panel (Workflow tier) — diverse generators, majority-vote consensus, cautious-confidence judge, top-K.
fierce-compete
Deterministic competitive code tournament (Workflow tier) — best of N implementations.
fierce-review
Deterministic adversarial code review (Workflow tier) for high-stakes scope.
fierce-debate
Deterministic adversarial debate (Workflow tier) for high-stakes decisions.