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/shivaedev/pardes/orchestratenpx skills add ShivaeDev/pardes --skill orchestrategit clone --depth 1 https://github.com/ShivaeDev/pardesWhat 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 | $0.00054 | $0.02344 |
| Opus 5 | $0.00027 | $0.01172 |
| Sonnet 5 | $0.00011 | $0.00469 |
| Haiku 4.5 | $0.00005 | $0.00234 |
Grade A, and why
orchestrate 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 2d 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 — 147 lines — stays where its author put it; the contents beside it link to each section on GitHub.
orchestrate
A two-phase workflow for big PRs that would otherwise overflow context or get bogged down mid-stream.
Phase 1 — Plan. A multi-round interview narrows scope and shape before any code. Phase 2 — Dispatch. Sequential sub-agents, one chunk per agent, one commit per chunk.
The orchestrator (you) stays high-level. Sub-agents do the coding. That separation is what unlocks big PRs in one session.
When to use
- The work spans more files than one agent can hold (typically: ≥30 files modified, multiple subsystems).
- The work has internal sequencing — chunk N depends on chunk N-1 in a way that makes parallelism counterproductive.
- The user has explicitly asked you to drive it autonomously to a PR, or it's obvious from the brief that's the shape.
Don't use for: single-file changes, exploration questions, work that fits in one agent context, or work where the sequencing isn't yet clear (plan first, dispatch later).
Phase 1 — Plan
Interview the user across multiple AskUserQuestion rounds
Bias toward more rounds, not fewer. Each round narrows one tier of decision. A representative arc:
- Scope tier. Bundled vs phased. Which adjacent work to absorb. What to defer.
- Shape tier. Interface shape, manifest shape, data-model shape. The structural decisions that fall out of scope.
- Detail tier. Specific field semantics, relationship directions, invariant rules. Anything the user hasn't seen articulated yet.
Use AskUserQuestion with 3–4 options per question, and mark the recommended one. If the user pushes back on a framing ("that concept should disappear entirely"), reframe and re-ask — don't proceed on a misread.
A pivot inside the interview is cheap. A pivot during execution is expensive. Spend the rounds.
Write the plan visibly before any code
After the interview settles, output two things in one message:
- Goals (ideal end state). Bulleted by area. What the world looks like after the PR merges.
- Sequence. Numbered chunks, one paragraph each. Each chunk specifies: what it changes, what it depends on, and whether it commits clean or as intentional WIP.
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.
- 2d ago First seen · 147 lines · 54 tokens per session scan A 1113cd655fb7
orchestrate is a skill published in the GitHub repository ShivaeDev/pardes (5 stars, last pushed 1mo ago), licensed MIT. It adds 54 tokens to every session and 2,344 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.
Other skills, from other repositories
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
next-cache-components-adoption
Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…
babysit-pr
Babysit a GitHub pull request after creation by continuously polling review comments, CI checks/workflow runs, and mergeability state until the PR is merged/closed or user help is required. Diagnose failures, retry likely flaky failures up to 3 times, auto-fix/push branch-related issues when appropriate, and keep…
imagegen
Generate or edit raster images when the task benefits from AI-created bitmap visuals such as photos, illustrations, textures, sprites, mockups, or transparent-background cutouts. Use when Codex should create a brand-new image, transform an existing image, or derive visual variants from references, and the output…
cpu-profile-analysis
Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…
next-cache-components-optimizer
Drive a Next.js route to instant navigation by setting up an agentic loop, under Cache Components / PPR, on initial load (hard navigation) and client-side navigation (soft navigation). Encode the goal as a failing @next/playwright instant() e2e and work it to green, one verified route at a time; the shipped test then…