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/ecro/harness-maker/makegit clone --depth 1 https://github.com/Ecro/harness-makerWhat 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 | $0.00000 | $0.10856 |
| Opus 5 | $0.00000 | $0.05428 |
| Sonnet 5 | $0.00000 | $0.02171 |
| Haiku 4.5 | $0.00000 | $0.01086 |
Grade A, and why
make 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 yesterday.
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 — 761 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/harness-maker:make
Refresh, reconfigure, or extend the project's harness at <cwd>/.claude/.
Asks the user what they want to do before invoking the CLI.
Procedure
You (Claude) act as the orchestrator. Follow these steps:
0. CI / test mode detection
If the prompt text contains --ci, extract inline params and skip all
AskQuestion (Cursor) / AskUserQuestion (Claude Code) calls. Parse preset=, locale=, dev_mode=,
targets= from the prompt; use defaults Side / en / task /
claude-code for any that are absent. Skip the live locale question and
jump directly to section 5 (Dispatch → Fresh install or Update,
depending on whether .claude/harness.yaml exists).
Example invocation:
/harness-maker:make --ci preset=Side locale=en dev_mode=task targets=claude-code,cursor,codex
0.5. --reinterview shortcut
If the prompt text contains --reinterview (and is not --ci), the user
wants a full interactive reconfigure. Do not pass --reinterview through
to the CLI — the CLI's interactive interview reads from stdin, which
falls back to autoloop defaults in the slash-command context (no TTY).
Instead, skip the menu in section 3 and jump straight to the Full
reconfigure branch in section 5, which drives AskQuestion (Cursor) / AskUserQuestion (Claude Code) here in
the slash command and dispatches with collected --preset / --locale / --dev-mode / --targets flags.
1. Choose live locale
Default is en, but the first interactive decision must be the user's live
onboarding language. Ask this before profile detection, preset selection,
dev_mode, targets, or any setup confirmation. All subsequent live onboarding prose,
question text, option labels, trade-off explanations, and decision
summaries must use the selected locale. Persisted generated documents may still
follow their own template rules; this section controls the live setup
conversation.
Use AskQuestion (Cursor) / AskUserQuestion (Claude Code):
- English (
en) — default; best-supported public-plugin baseline. - Korean (
ko) — use Korean for the rest of this onboarding flow. - Other locale tag — accept a free-text BCP-47-style tag; if no localized template exists, explain that runtime generated text may fall back to English.
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.
- yesterday First seen · 761 lines · 0 tokens per session scan A fc2f6cee6996
make is a command published in the GitHub repository Ecro/harness-maker (4 stars, last pushed 6d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 10,856 tokens. 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 commands, from other repositories
constitution
根据交互输入或已提供原则创建或更新项目章程，并确保相关模板保持同步。.
meta-theory-verify
Run the appropriate MetaKim verification path.
cross-research
Two-phase multi-model research. Phase A fans codex+gemini in parallel (benchmarks + citations angles); Phase B synthesises via fresh Claude session. Usage: /cross-research [--with | list | compare].
ijfw-audit
Run the IJFW audit gate for the current workflow phase. Usage: /ijfw-audit [phase name].
memory-why
Show why a memory recall returned what it did -- BM25 vs vector vs hybrid provenance.
ijfw-execute
Jump directly to the IJFW workflow Execute phase (Deep D4 / Quick Q3). Usage: /ijfw-execute [task or phase name].