Memory & Context

Find ways to cut AI coding spend from your own session logs, with savings estimates

Flag unusually expensive sessions using median-based outlier scores for CI gates

Run a corpus benchmark and save a verifiable table of measured versus claimed cost

Apply simple code transforms in a WASM engine with no LLM call and zero cost

Route tasks to zero-cost code transforms when a booster can handle them

Check spend against your budget and alert at 50, 75, 90 and 100 percent

Track spend burn rate over time and alert when the latest bucket spikes

Pull compacted context for cost queries and report the tokens it saved

List each session with its start time, message count, top model and cost

Test whether model routing saves money against always using one fixed tier

Diff two cost summary snapshots to catch spend regressions in a pull request

Track spend per federation peer and flag peers that cross a suspension threshold

Run four cost checks in one CI gate and return a single health status

See spend, tokens and cache hit rates across AI coding tools in one local ledger

Check OpenRouter spend, key limits and credit balance across day, week and month

Analyze token usage patterns and recommend changes that cut spend

Project spend over 7, 30, 90 and 365 days and estimate when budget runs out

Report token use and USD cost broken down by agent and model

Break down the cost of one session message by message to find costly turns

Dump total spend, tier breakdown and budget status as JSON or markdown

Capture token usage from session logs automatically and store it for cost tracking

Track benchmark drift in win rate, latency and cost across saved runs

Copy learned patterns between projects or share them through a pattern marketplace

Bridge coding tool memory into AgentDB with embeddings and cross-project search

Search Ruflo memory with semantic, keyword or reranked retrieval

Store approved team decisions and facts with provenance and search them later

Manage portable RVF memory files for moving agent memory across platforms

Save and restore agent sessions across conversations with state snapshots

Store and search long-term agent memories with Mem0 through Composio Rube MCP

Search memories from earlier Pi sessions before repeating reads and experiments

Delete Mem0 memories stored for this repository and user on request

Add persistent long-term memory to chatbots and AI apps with the Mem0 SDK

Run Mem0 add, search, list and get commands from the terminal with the CLI

Search memories from earlier OpenCode sessions before repeating file reads

Delete Mem0 memories found by search, after confirming each removal

Integrate Mem0 into an existing repo test-first on a feature branch

Plan and run a migration from self-hosted Mem0 to the hosted Mem0 Platform SDK

Store a remember-this note in Mem0 verbatim, with its type and metadata

View or change whether Mem0 memories are scoped to project, session or global

Look up Mem0 memories in compact one-line results or fetch one by ID

Diagnose Mem0 connectivity, API key validity and failing memory reads or writes

Browse all stored project memories grouped by category with full content

Decide which durable facts, preferences and decisions to store in long-term memory

Pause Mem0 memory capture on this machine until you resume it

Save a remember-this request as a durable memory for future sessions

Turn Mem0 memory capture back on after it was paused

Search memories from earlier sessions to avoid repeating reads and experiments

Check that Mem0 memory is working here, including config and API key validity

Browse every stored memory grouped by category to get an overview