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 skills/emtcmca/promptsmith/orchestrationnpx skills add emtcmca/promptsmith --skill orchestrationgit clone --depth 1 https://github.com/emtcmca/promptsmithWrote 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/emtcmca/promptsmith/orchestration)<a href="https://agentmods.dev/skills/emtcmca/promptsmith/orchestration"><img src="https://agentmods.dev/badge/skills/emtcmca/promptsmith/orchestration.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 5.1 | $0.00109 | $0.03928 |
| Opus 5 | $0.00055 | $0.01964 |
| Sonnet 5 | $0.00022 | $0.00786 |
| Haiku 4.5 | $0.00011 | $0.00393 |
Grade C, and why
orchestration scanned grade C 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.
Harvests environment variableshighData exfiltration
Enumerating or grepping the environment for keys collects credentials unrelated to what the mod says it does.
harvest credentials, impersonate a person/institution, surveil without consent, build How it starts
The opening of the file, as written. The whole thing — 272 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Orchestration — the promptsmith coordinator (Layer 2)
This is the coordinator. It takes a multi-domain request, perfects it, splits it across the gallery specialists, and assembles their work into one deliverable — owning the seams the individual agents can't see.
Layer boundary. Unlike the Layer 1 core (/sharpen, /forge-agent, /lens — zero model
calls, paste-anywhere), this layer dispatches live subagents and so requires a host that
can spawn them (Claude Code's Agent/Task capability). It is still method + structure: the host
runs this protocol; there is no bundled runtime.
When to use
- A request clearly spans more than one specialist's purview (e.g. "build a password reset flow" → spec + data + security + API + UI + tests + docs).
- The value is in coverage and coherence — many slices that must end as one deliverable.
When NOT to use (fall back — Step 0): a single agent or plain /sharpen handles it well.
Orchestration has real cost (fan-out + synthesis); don't pay it for a one-domain task.
Context discipline — keep big things out of your window
The coordinator must stay context-aware: hold the plan, the seams, and the conclusions — not the raw material. Subagent context is disposable; exploit that. A coordinator that pulls large artifacts into its own window goes blind faster and loses the thread.
- Never read a large artifact into your own context to review, edit, or search it — a big file, a whole directory, a long log, a corpus, a giant tool output. Dispatch a subagent that reads it in its context and returns only the distilled result: findings, a diff, a verdict, a summary, the specific lines.
- Pass references, not contents. Hand subagents file paths, globs, line ranges, IDs — let them fetch. Don't paste a file's body into a slice goal.
- Editing a large file: dispatch a builder to produce or apply the change in its context and
return a compact diff /
--stat; review the diff, not the file. (If you already have a subagent that produced the change, resume it to apply — it still holds the context.) - Synthesize from distilled returns, never by re-ingesting every slice's raw output. If a single return is itself large, route it through a summarizer before you hold it.
- The test: if reading something would burn a big chunk of your window and you only need the conclusion, that's a subagent's job, not yours.
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 · 272 lines · 109 tokens per session scan C 793013731b17
orchestration is a skill published in the GitHub repository emtcmca/promptsmith (2 stars, last pushed 2d ago), licensed Apache-2.0. It adds 109 tokens to every session and 3,928 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it C with 1 finding (harvests environment variables). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
improve
Autonomous quality improvement loop. Scores a target against a rubric, selects the highest-leverage axis, attacks it, verifies, documents, and loops. No pre-planning between iterations — each loop re-scores from scratch.
evolve
Research-driven multi-cycle improvement director. Forms causal hypotheses about why scores are low, validates them with scout agents before attacking, dispatches axis-parallel fleet attacks, extracts transferable patterns, and runs indefinitely within a budget envelope. Accumulates a persistent belief model and…
research
Focused research investigations. Converts questions into structured findings with confidence levels and source citations. Single agent by default; with --parallel (or when the question decomposes into 3+ independent angles) it spawns scout agents whose findings are compressed into a unified brief. Does not make…
pr-watch
Local PR watcher. Monitors CI status, automatically fixes failing checks by reading failure logs and applying targeted fixes, then optionally merges when all checks pass. Local CLI analog to Claude Code's cloud auto-fix feature.
qa
Browser-based QA verification. Launches a real browser, navigates the app, clicks buttons, fills forms, and tests user flows. Works as a standalone skill or as a phase end condition in campaigns. Requires Playwright (optional dependency, graceful skip if not installed).
live-preview
Mid-build visual verification loop. Takes screenshots of components during construction, not just after. Catches visual regressions and invisible features before they compound. Requires Playwright or similar screenshot tool.