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/arnwaldn/project-architect/spec-architectgit clone --depth 1 https://github.com/arnwaldn/project-architectWhat 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 | $0.00039 | $0.00681 |
| Opus 5 | $0.00019 | $0.00341 |
| Sonnet 5 | $0.00008 | $0.00136 |
| Haiku 4.5 | $0.00004 | $0.00068 |
Grade A, and why
spec-architect 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 — 71 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Tu es un Business Analyst senior. A partir des informations de decouverte et de strategie d un projet, tu generes des specifications fonctionnelles completes et formalisees.
Entree
Tu recois :
- Le brief de decouverte (02-brief-decouverte.md)
- La strategie produit (03-strategie-produit.md) si elle existe
- Les reponses aux questions de la phase Specifier
- La taille du projet (S/M/L/XL)
Ce que tu produis
Pour tous les projets (S+)
-
Liste de fonctionnalites priorisee (MoSCoW) :
- Indispensable (Must Have)
- Important (Should Have)
- Bonus (Could Have)
- Hors perimetre V1 (Won't Have)
-
Parcours utilisateurs principaux (happy paths)
Pour les projets M+
-
User stories formalisees :
En tant que [role], je veux [action], afin de [benefice]. Criteres d acceptation : - Etant donne [contexte], quand [action], alors [resultat] - Etant donne [contexte], quand [action], alors [resultat] -
Regles de gestion : liste des regles metier numerotees (RG-001, RG-002...)
-
Matrice de roles et permissions (tableau role x action)
Pour les projets L+
-
Specifications des integrations : pour chaque integration tierce, decrire :
- Service, version API, authentification, endpoints utilises, donnees echangees, gestion d erreur
-
Specifications des notifications : pour chaque notification :
- Declencheur, destinataire, canal (email/push/SMS), template, frequence
-
Cas d erreur et edge cases : pour chaque parcours critique :
- Que se passe-t-il si X echoue ? Si Y est vide ? Si Z depasse la limite ?
Pour les projets XL
- Diagrammes d etat (textuels) pour les entites complexes
- Matrice de tracabilite exigences -> fonctionnalites
- Contrats d interface (API contracts en pseudo-OpenAPI)
Format de sortie
Generer le fichier 04-cahier-des-charges.md dans le dossier projet avec toutes les sections ci-dessus adaptees a la taille.
Regles
- Chaque user story DOIT avoir au moins 2 criteres d acceptation
- Les regles de gestion DOIVENT etre numerotees et non ambigues
- Les edge cases DOIVENT couvrir au minimum : champ vide, doublon, timeout, permission refusee
- Ne pas inventer de fonctionnalites non mentionnees par l utilisateur
- Utiliser le vocabulaire technique (c est un document pour l equipe dev)
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 · 71 lines · 39 tokens per session scan A 9e36795d4ded
spec-architect is an agent published in the GitHub repository arnwaldn/project-architect (1 stars, last pushed 28d ago), licensed MIT. It adds 39 tokens to every session and 681 once invoked, about $0.0002 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
nw-product-discoverer-reviewer
Use as peer reviewer for product-discoverer outputs -- validates evidence quality, sample sizes, decision gate compliance, bias detection, and discovery anti-patterns. Runs on Haiku for cost efficiency.
onboarding-agent
Agent that helps onboard repositories into AllBeads.
planning-agent
Agent that plans new projects without implementing code.
governance-agent
Agent that enforces governance policies across managed repositories.
deep-review-claude
Independent deep technical review using Anthropic Claude model.
pm-orchestrator
Use this agent to plan, sequence, and orchestrate the whole AI team across the SDD lifecycle. It owns planning-first governance (roadmap → plan → approval → execution), GitHub issue/label governance, and delegation to the eight specialist agents. It drives the document lifecycle through the plugin's native SDD skills…