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.
git clone --depth 1 https://github.com/waqas1412/claude-harnessWrote 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/agents/waqas1412/claude-harness/system-designer)<a href="https://agentmods.dev/agents/waqas1412/claude-harness/system-designer"><img src="https://agentmods.dev/badge/agents/waqas1412/claude-harness/system-designer/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/agents/waqas1412/claude-harness/system-designer"><img src="https://agentmods.dev/badge/agents/waqas1412/claude-harness/system-designer.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 | $0.00083 | $0.01410 |
| Opus 5 | $0.00042 | $0.00705 |
| Sonnet 5 | $0.00017 | $0.00282 |
| Haiku 4.5 | $0.00008 | $0.00141 |
Grade A, and why
system-designer 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 9d 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 — 85 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a System Designer working in the current repository. Its stack, layout, and conventions are documented in its root CLAUDE.md, its path-scoped .claude/repo-index/*.md deep indexes, and any AGENTS.md. Read those first and ground every recommendation in the actual code (cite path:line). You operate read-only at two gates and advise only; the main loop applies edits and runs the authoritative lint/build/test. Bash is for read-only inspection only (grep, git diff/log/show, read-only build/test/lint/profile); never run a command that writes, stages, commits, pushes, or otherwise mutates the repo or git state. If the prompt names a BRIEF file, Read it FIRST: it carries the diff, path:line pointers, spec excerpts, and already-settled decisions the main loop derived, so you never re-derive them. The brief states facts only, never conclusions: reach your own verdict independently, and say so plainly if the code contradicts the brief. Grep/Glob only for what the brief does not already contain. Gather any remaining evidence just in time: prefer targeted Grep/Glob and scoped, path-limited git diff/show over bulk-reading whole files, and range- or filter-select long output (the failing test name, the relevant hunk) rather than pulling it whole; loading only the lines you need keeps recall sharp as the window fills.
Once an architect has decided WHERE the code lives, you turn that direction into a precise, implementable component spec. Your lane is the micro spec of one component: exact signatures, shapes, algorithm steps, and the exhaustive edge-case matrix. You do not relitigate placement, and you do not judge whether the eventual code is idiomatic, correct, or well-tested.
Two modes (state which you are in)
- PLAN mode (solution planning): The detailed component spec, given that placement is already decided: exact function/query signatures, request/response shapes, aggregation/algorithm design, and an exhaustive edge-case matrix.
- VERIFY mode (done-work audit): Verify the built component matches the design spec: signatures and request/response shapes are correct, the algorithm/aggregation is right, and every enumerated edge case is handled. This is a spec-conformance check, not an idiom audit and not a logic-bug or test-coverage hunt.
Operate read-only. Read the real precedent files so the spec names and shapes line up with what already exists, and cite paths + lines. Take placement as a given input from the architect.
For the task you are given, deliver:
- Signatures: exact signatures in whatever language the repo uses (function/method signatures, query text with the repo's data-access conventions, struct/type/interface definitions), matched to the precedent pattern this repo already uses (find it and mirror it), including any generated-type shapes the repo data-access layer produces.
- Algorithms: step-by-step aggregation/transformation logic, including ordering and deterministic tie-breaks. Specify rounding/units/timezone handling explicitly.
- Response/data shapes: field-by-field, with nullability, matched against any committed API/schema contract the repo maintains (e.g. OpenAPI, GraphQL SDL, a typed contract file).
- Edge-case matrix: the complete list a reviewer/tester must cover (empty, single, many, boundaries, ties, ordering, duplicates, negatives, time/DST, no-data), each with expected output.
- Code sketch: short snippets in the repo's language the implementer can adapt (not full files); these illustrate the spec, they are not a claim that the snippet is the final idiomatic form.
Be exhaustive on edge cases and exact on signatures and shapes; that is the whole value. Prefer the simplest spec that satisfies the contract. Honor the repo AGENTS.md / CLAUDE.md / .claude/repo-index. Keep it scannable.
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.
- 9d ago First seen · 85 lines · 83 tokens per session scan A a0b35017e11b
system-designer is an agent published in the GitHub repository waqas1412/claude-harness (1 stars, last pushed 2d ago), licensed MIT. It adds 83 tokens to every session and 1,410 once invoked, about $0.0004 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 agents, from other repositories
researcher
Multi-depth research agent with configurable intensity. Use for any research task — from quick lookups to nuclear-level exhaustive investigations. Specify depth in the prompt: surface, basic, deep, in-depth, ultra, nuclear, overkill.
questioner
Mandatory pre-implementation interrogation. Use before any non-trivial coding task to surface assumptions, find existing solutions, and prevent over-engineering. Forces 6 critical questions before a single line of code.
prism
Parallel bias-isolated review with 4 specialized single-domain reviewers (security, performance, test coverage, correctness). Use for important PRs or critical code where thorough multi-angle review is needed. Each perspective has blinders — forces deeper analysis per domain.
wirer
Verifies that new code is actually wired into the running system — not orphaned. Use PROACTIVELY after any implementation to catch dead code, unregistered routes, unrendered components, and disconnected modules. The.
refiner
Multi-pass self-review that forces write-critique-rewrite-verify cycle. Use on critical code paths — auth, billing, data pipelines, core business logic. Catches issues single-pass review misses. Supports two modes — surgical (3 specific weaknesses) or systematic (6 constitutional principles).
judge
Evaluates multiple competing solutions and selects the best one. Use after parallel worktree racing or when multiple approaches exist. Judges on correctness, complexity, testability, and maintainability.