opscompanion-remember

OpsCompanion Remember

You are saving a decision, discovery, or piece of context to the team's shared knowledge base so it can be recalled in future sessions — by this user or any teammate.

Action

Run:

opc --agent claude remember "<content>" --tags "<tag1>,<tag2>,<tag3>"

Crafting Good Memories

Content should be:

• Self-contained — readable without surrounding context
• Specific — include the "what" and the "why"
• Actionable — someone finding this later should know what to do

Bad	Good
"use Redis"	"Use Redis for rate limiter shared state — in-memory counters don't work across multiple gateway pods"
"PASETO is better"	"Migrating auth-service from JWT to PASETO v4. PASETO chosen for mandatory encryption and no algorithm confusion attacks"

Tags should include:

• The service name (e.g., api-gateway, auth-service)
• The topic area (e.g., rate-limiting, security, architecture)
• The type of knowledge (e.g., debugging, migration, config)

When to Proactively Remember

Save a memory when you observe:

• A significant architectural decision being made
• A non-obvious discovery during debugging
• A decision with context that would be lost after the session
• Something the user explicitly says "remember this" or "we should document this"

Always confirm with the user before proactively saving.

If the user is still investigating a live issue, gather the runtime evidence first with opscompanion-observability and only save the resulting conclusion once it is clear.

Response Format

After saving:

• Confirm what was saved (paraphrase, don't repeat verbatim)
• Show the tags assigned
• Mention that teammates can find this with opc search (from any agent)

Error Handling

If opc --agent claude remember fails with a config error, invoke the opscompanion-init skill to set up the CLI first.