Installable local agent
Runs on your machine, close to the work.
Features
eri installs as a local agent and terminal console, then connects MCP tools, APIs, skills, docs, datasets, traces, and cloud runners into one previewable loop: read, plan, build, review, checkpoint, and ship.
The full catalog
Local agents, terminal control, MCP tools, universal source reading, generated interfaces, review checkpoints, cloud handoff, and voice or gaze when hands-free context matters.
Local first
01 / 05
Runs on your machine, close to the work.
Ask, run, inspect, and recover in shell.
See active agents, queues, and handoffs.
Send long jobs to managed background agents.
Bring external tools with your own keys.
Teach repeatable workflows once, reuse them.
Universal context
02 / 05
Search files, symbols, diffs, and history.
Find across local, web, docs, and tools.
Summarize, compare, and cite project docs.
Attach tables, evals, traces, and examples.
Follow decisions from prompt to command.
Keep watched repos and docs fresh.
Move context between local and cloud agents.
Control what leaves your machine.
From ask to ship
03 / 05
Review scope before an agent touches files.
Plan, edit, run, verify, and explain.
Read diffs, risk, logs, and test output.
Every turn can be inspected or rolled back.
Commit, branch, push, and open PRs.
Commands and writes stay gated by consent.
Eri GenUI
04 / 05
Agents create task-specific interfaces.
Inspect runnable outputs before applying.
Keep generated UIs, docs, and patches.
Watch changes resolve into reviewable diffs.
Hands free when useful
05 / 05
Talk through local work without leaving flow.
Look at the thing; eri resolves context.
Drop screenshots, mocks, and diagrams.
Pick local or hosted models by task.
Next
Features are easier to feel than to read. The demo walks the local workbench, agent manager, context graph, generated UI, and ship loop in under three minutes.