continual-learning

par microsoft

Guide pour implémenter l'apprentissage continu dans les agents de codage IA — hooks, cadrage mémoire, motifs de réflexion. À utiliser lors de la mise en place d'une infrastructure d'apprentissage pour…

npx skills add https://github.com/microsoft/skills --skill continual-learning

Télécharger ZIP GitHub

2.9k

Continual Learning for AI Coding Agents

Your agent forgets everything between sessions. Continual learning fixes that.

The Loop

Experience → Capture → Reflect → Persist → Apply
     ↑                                       │
     └───────────────────────────────────────┘

Quick Start

Install the hook (one step):

cp -r hooks/continual-learning .github/hooks/

Auto-initializes on first session. No config needed.

Two-Tier Memory

Global (~/.copilot/learnings.db) — follows you across all projects:

Tool patterns (which tools fail, which work)
Cross-project conventions
General coding preferences

Local (.copilot-memory/learnings.db) — stays with this repo:

Project-specific conventions
Common mistakes for this codebase
Team preferences

How Learnings Get Stored

Automatic (via hooks)

The hook observes tool outcomes and detects failure patterns:

Session 1: bash tool fails 4 times → learning stored: "bash frequently fails"
Session 2: hook surfaces that learning at start → agent adjusts approach

Agent-native (via store_memory / SQL)

The agent can write learnings directly:

INSERT INTO learnings (scope, category, content, source)
VALUES ('local', 'convention', 'This project uses Result<T> not exceptions', 'user_correction');

Categories: pattern, mistake, preference, tool_insight

Manual (memory files)

For human-readable, version-controlled knowledge:

# .copilot-memory/conventions.md
- Use DefaultAzureCredential for all Azure auth
- Parameter is semantic_configuration_name=, not semantic_configuration=

Compaction

Learnings decay over time:

Entries older than 60 days with low hit count are pruned
High-value learnings (frequently referenced) persist indefinitely
Tool logs are pruned after 7 days

This prevents unbounded growth while preserving what matters.

Best Practices

One step to install — if it takes more than cp -r, it won't get adopted
Scope correctly — global for tool patterns, local for project conventions
Be specific — "Use semantic_configuration_name=" beats "use the right parameter"
Let it compound — small improvements per session create exponential gains over weeks