trajectory

name	trajectory
description	Trajectory Learning for Botte Secrète — stores solver trajectories and searches similar past optimizations to inform future decisions
version	1.0.0
author	Hermes Agent
license	MIT
metadata	{"hermes":{"tags":["botte","trajectory","memory","learning"],"related_skills":["solvers","trajectory-memory"]}}

Trajectory Learning for Botte Secrète

Overview

Stocke les trajectoires d'optimisation des solveurs déterministes. Permet de retrouver des solutions similaires déjà calculées → 0 token, 0 latence.

Compatible avec ~/.hermes/scripts/trajectory_search.py (Hermes Trajectory Memory).

Files

Fichier	Rôle
skills/trajectory/__init__.py	capture(), search(), load(), stats()
skills/trajectory/store/trajectories.jsonl	Stockage JSON Lines

Usage

from skills.trajectory import capture, search, get_stats

# Capturer une trajectoire
tid = capture(
    solver="bin_pack",
    task="pack 3 database backups into 10GB capacity",
    parameters={"items": [("db1", 4), ("db2", 7), ("db3", 3)], "capacity": 10},
    result={"bins": [...], "bin_count": 2},
    latency=0.002,
    tokens_saved=500,
)

# Rechercher des trajectoires similaires
hits = search("pack items into capacity")
for h in hits:
    print(f"[{h['score']:.2f}] {h['trajectory']['task']}")

# Statistiques
stats = get_stats()
print(f"Total: {stats['total']}, Tokens saved: {stats['total_tokens_saved']}")

Integration with solvers

Dans skills/solvers/solvers.py, chaque fonction peut capturer sa trajectoire:

from skills.trajectory import capture

def assign_balanced(tasks, workers):
    result = ...  # existing logic
    capture("assign_balanced", f"assign {len(tasks)} tasks to {len(workers)} workers",
            {"tasks": tasks, "workers": workers}, result, latency=latency())
    return result

Integration with Hermes

Les fichiers sont compatibles: Botte peut lire les trajectoires Hermes et vice-versa:

# Lire les trajectoires Hermes
from skills.trajectory import load
hermes_trajs = load("/home/redgamer/.hermes/trajectory_store/trajectories.jsonl")

Pitfalls

• Rotation automatique à 5000 entrées
• TF-IDF uniquement — pas de sémantique profonde
• Latence de capture < 1ms (append-only)
• Thread-safe pour usage concurrent