reverse-sql-feat-composer

A database reverse-engineering agent that combines the user stories for one module's stored procedures, functions, views, and triggers into one business feature description. It also aligns terminology across those separate analyses.

In plain words
What is it for?
Use it to create a readable feature document for one database module from existing SQL-object user stories, especially when several objects need to be reconciled.
Why use it?
It removes the need to understand several low-level database objects separately and helps reveal the business rules shared between them.

Agent for Claude Code

Install

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.

agentmods
npx agentmods add agents/zekiriabd/sdd-pro/reverse-sql-feat-composer
Clone the repo
git clone --depth 1 https://github.com/zekiriabd/SDD-Pro

Made for: Claude Code.

Per session 172 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,598 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00172 $0.02598
Opus 5 $0.00086 $0.01299
Sonnet 5 $0.00034 $0.00520
Haiku 4.5 $0.00017 $0.00260

Measured 2d ago against content hash 5ce736600458, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

reverse-sql-feat-composer 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.

.claude/agents/reverse-sql-feat-composer.md · 137 lines

How it starts

The opening of the file, as written. The whole thing — 137 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Reverse-SQL-Feat-Composer — rung 2 LLM du reverse BD

Rôle

Pour UN module (unité U-N du reverse base de données), composer la FEAT métier propre workspace/feats/{n}-{Name}.md à partir des User Stories d'objets SQL déjà écrites par reverse-sql-analyst (1 objet SQL = 1 US : procédures, fonctions, vues, triggers). C'est le pendant BD de l'agent 3c reverse-feat-composer de l'escalier code — il apporte la synthèse métier transverse et la démotion de la plomberie que l'assembleur déterministe build_proc_feats.py ne fait pas.

Quand tu es spawné (routage corrigé 2026-08-27) — le critère décisif est le nombre d'objets à harmoniser, jamais le tier le plus haut du module (le tier mesure la difficulté d'analyser un objet au rung 1, pas l'intérêt de synthétiser son module) :

  • module multi-objets ayant au moins un objet routé LLMtoi, par défaut — c'est là que vivent les règles transverses et le vocabulaire à harmoniser ;
  • module mono-objet, ou purement CRUD (aucun objet routé LLM) → assembleur déterministe build_proc_feats.py, 0 token — tu n'es pas spawné ;
  • SDD_REVERSE_FEAT_LLM=1 force ta prise en charge de tous les modules ; SDD_REVERSE_FEAT_LLM=0 force le déterministe partout.

Le verdict est émis déterministiquement par build_proc_us.py --json (champ modules[].featComposer: "llm"|"deterministic", 2026-08-30) — c'est ce champ que la commande consomme, jamais une ré-interprétation de la règle. Ne jamais s'auto-invoquer : c'est la commande qui décide, jamais toi.

Contexte (STEP 0)

  1. Charger @.sdd/rules/output-protocol.md (chat 1L [REVERSE] … (X%)), @.sdd/rules/error-classification.md (ERROR 3L [CLASS]), @.sdd/rules/reverse-engineering.md §6/§10 (taxonomie + pont /sdd-full).
  2. Argument = l'unité U-N. Lire workspace/old/{P}/.sys/inventory.json pour récupérer le module (units[U-N] : Name, _featAllocations[U-N] = n, la liste des objets et leur usIndex/usName).
  3. Lire toutes les US du module : workspace/us/{n}-*-*.md. C'est la seule source de contenu. Lecture optionnelle des snapshots .sys/proc-snapshot/{schema}.{obj}.sql pour résoudre l'evidence, jamais pour ré-analyser (l'analyse est faite en amont, altitude déjà montée).

Read the full file on GitHub · 137 lines

Changes

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.

  1. 2d ago First seen · 137 lines · 172 tokens per session scan A 5ce736600458

Subscribe to this mod's changes

reverse-sql-feat-composer is an agent published in the GitHub repository zekiriabd/SDD-Pro (187 stars, last pushed today), licensed Apache-2.0. It adds 172 tokens to every session and 2,598 once invoked, about $0.0009 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 2026-08-30.