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/langerrr/zforge/reviewgit clone --depth 1 https://github.com/Langerrr/zforgeWhat 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 | $0.00006 | $0.01030 |
| Opus 5 | $0.00003 | $0.00515 |
| Sonnet 5 | $0.00001 | $0.00206 |
| Haiku 4.5 | $0.00001 | $0.00103 |
Grade A, and why
review 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 — 68 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/review — Ledger vs. Implementation
Check what the code does against what the feature's documents say it does.
This is not a general code review. Bugs, security, performance and style belong to /code-review and ce:review, which are better at them and do not need the ledger. This review is the one nobody else can run, because it is the only one holding the plan, the patterns, the decisions and the evidence claims at the same time.
Arguments
--feature <name>— the feature directory to review against. Required; without a ledger there is nothing here to do.
What to read first
02_plan.md— schema, contracts, phase decomposition, Verification Matrix, Environment Assumptions, Invariants- The core patterns doc, if the Doc Map lists one — the rules agents were told to follow
decision_review.md§A and §C — the decisions that reached beyond the implementation, and whatever the user has already adjudicated- Every phase file's
## Decisions— the implementation conventions that were not promoted. This is the larger set, and it binds the code just as tightly - Every phase file's
## Evidence Requiredand## Acceptance 05_progress_overview.md— open standing flags
Then read the code those documents describe.
What to check
Four questions, each answerable against a document rather than a judgment call:
1. Do the decisions hold in the code?
Every 🟡 and ✅ entry in decision_review.md, and every row in a phase file's ## Decisions, claims something about how the system works. Check the call sites. A decision recorded once and violated in four places is the common shape, and it is commonest among the unpromoted conventions — the ledger's rows had the user's eye on them, the phase files' rows had nobody's.
2. Are the invariants implemented and re-checked?
02_plan.md names invariants with an owner phase and a re-check phase. Verify each is actually established, and actually holds at every call site — not just the one the owning phase wrote. Anything spanning phases is verified by no single phase file, which is why it survives to here.
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 · 68 lines · 6 tokens per session scan A 5e4435ee5f83
review is a command published in the GitHub repository Langerrr/zforge (10 stars, last pushed 2d ago), licensed MIT. It adds 6 tokens to every session and 1,030 once invoked, about $0.0000 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 commands, from other repositories
decision
Record one architecture decision in the decision log (what, why, what was rejected, when to revisit) / Registra una decisión de arquitectura en la bitácora de decisiones.
tutor
Learn SDD conversationally - the agent becomes your tutor, by levels, with real validation as the grader / Aprende SDD conversando - el agente es tu tutor, por niveles, con validación real como corrector.
gate
Run the SDD gate (policy + approval + consistency) and record consent before implementation / Ejecuta la compuerta SDD y registra consentimiento antes de implementar.
close
Validate and close the session with the SDD output contract / Valida y cierra la sesión con el contrato de salida SDD.
new
Start a project with SDD - capture the idea and prepare the first spec / Inicia un proyecto con SDD - captura la idea y prepara la primera spec.
serve
Serve the SDD Builder AI request queue - draft proposals, never write specs / Atiende la cola de peticiones del SDD Builder - propone, nunca escribe specs.