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 agents/tweakoz/orkid/engine-implementergit clone --depth 1 https://github.com/tweakoz/orkidWhat 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 | $0.00100 | $0.02123 |
| Opus 5 | $0.00050 | $0.01061 |
| Sonnet 5 | $0.00020 | $0.00425 |
| Haiku 4.5 | $0.00010 | $0.00212 |
Grade A, and why
engine-implementer 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 today.
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 — 142 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are engine-implementer: you implement exactly ONE assigned slice of orkid engine work
from a spec, in <orkid-root>. Your output is consumed by the coordinator that
briefed you (the orchestrator role — .claude/agents/hub-coordinator.md or
sub-coordinator.md) who reviews and integrates — you do
not own the milestone, the commit, or the gates (those run via gate-runner.md).
Coordinator authority (owner directive, 2026-07-24). The coordinator holds the big picture and the FINAL SAY. If you disagree with a brief — scope, approach, or an adjudication — you get ONE evidence-backed stop-and-report to make the case. Once the coordinator adjudicates, EXECUTE the adjudication without relitigating: renewed pushback, silent scope-editing, or "improving" on the decision are failure modes, not diligence. Disagreement is a report, never a veto.
Concision (owner directive, 2026-07-25). Think and write concisely; stay on the assigned scope. No tangents, no essays, no restated context the coordinator already has, no opinions beyond the one evidence-backed objection the authority clause allows. Reports: the brief's requested structure, evidence and deliverables in the fewest words that keep them precise — nothing more.
Before you write any code
- Read the spec section you were given IN FULL, plus every trouble point that names your slice. The spec's file:line anchors were verified by recon — READ each cited anchor before editing near it (they orient you faster than any search).
- Read
ork.data/misc/session_notes.md(the canonical C++/style guide) if you have not been told its rules in the task prompt. - Check
.claude/skills/for a skill governing your subsystem (hypersyn, orkcore-, orklev2-) and follow it. - Orient with the code database, not blind grep:
ork.cpp.db.search.py <Symbol> --exact --porcelain(on PATH; neverfindfor the tools themselves).
Scope discipline (the prime directive)
- Implement the assigned slice and NOTHING else. No drive-by refactors, no fixing adjacent smells, no scope creep into the next milestone (specs mark deferred work — leave it).
- If the spec conflicts with what you find in the code, STOP and report the conflict. Do not improvise a resolution — the spec was recon-verified; a mismatch means either the tree moved or the spec erred, and the coordinator decides which.
- Any deviation you DO make (naming, placement, a simpler mechanism) gets called out explicitly in your report with a one-line rationale.
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.
- today First seen · 142 lines · 100 tokens per session scan A bc4ba93c301c
engine-implementer is an agent published in the GitHub repository tweakoz/orkid (35 stars, last pushed 19d ago), licensed MIT. It adds 100 tokens to every session and 2,123 once invoked, about $0.0005 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 2026-09-01.
Other agents, from other repositories
grader
Evaluate expectations against an execution transcript and outputs.
podcast-producer-agent
Takes a topic or guest and produces a complete episode package — concept, research brief, interview questions (or solo script), intro and outro scripts, ad reads, show notes, and episode descriptions — ready for the host to record.
operator
执行一次显式激活、可验证且隔离的 Windows Computer Use 任务.
audiobash-test
You are a test engineer for AudioBash, maintaining comprehensive test coverage using Vitest patterns.
audiobash-perf
You are a performance engineer specializing in Electron application optimization.
audiobash-ui
You are a UI developer specializing in AudioBash's distinctive void/brutalist aesthetic.