Alan for Teams

Fast agents. Faster teams.

better. faster. TOGETHER.

Share agent sessions with your teammates, keep every change attributed, search what the whole team has already learned, and catch duplicate work before it ships twice.

Sessions /Add rate limiting to the public API

Maya

Token bucket per API key. Dev, can you take the Redis side?

Claude Code · for Maya

Added the middleware and wired it into the gateway.

Edit middleware/rateLimit.ts

Dev · joined as co-author

On it. Codex, use the Lua script from the billing service so it stays atomic.

Found in team memory: Payments team solved Redis atomicity in “Billing retry limits”, 3 weeks ago.

Share with team

Works across every agent your team already uses: Claude Code · Codex · Cursor · opencode · GitHub · Slack · Linear

01 / Share

Pull a teammate into the session. On your terms.

Every session can be shared three ways. Pick how much of the work, and the conversation behind it, your team sees.

Co-author

Teammates join live. They prompt agents, steer the plan and edit the diff alongside you.

Prompt agents
Yes
Edit changes
Yes
See transcript
Yes

Read-only

The full transcript, diffs and checks, for reviewers, leads and anyone learning how the work got done.

Prompt agents
No
Edit changes
No
See transcript
Yes

Privacy mode

Share the outcome, not the conversation. Teammates get a summary of what was done and why. Your transcript stays yours.

Session summary

Added per-key rate limiting at the gateway. Chose token bucket over fixed window for burst tolerance. 2 files, checks passed.

02 / Attribution

Every change knows who made it.

Shared sessions are co-authored automatically. Each hunk carries the teammate who asked for it and the agent that wrote it, all the way into the commit and the pull request.

  • Per-line attribution for people and agents
  • Co-authored-by trailers written for you
  • Reviewers see who to ask about each decision
middleware/rateLimit.ts+24 −3

Maya · Claude Code

+ export function rateLimit(opts: Limits) {
+   const bucket = createBucket(opts.perKey);
+   return (req, res, next) => bucket.take(req.key) ? next() : res.status(429).end();
+ }

Dev · Codex

+ const take = redis.defineCommand(TOKEN_BUCKET_LUA);
+ // atomic: reuse billing service script

feat(api): per-key rate limiting

Co-authored-by: Maya S <maya@acme.dev>
Co-authored-by: Dev K <dev@acme.dev>
Agent-Session: alan://s/rate-limit-api

03 / Team memory

Your team already solved this. Now you can find it.

Search across every team’s sessions for previous work, artifacts, context and learnings. Reuse a teammate’s session, artifact or fix in your own work instead of rebuilding it.

04 / Duplicate work

Two sessions, same work. Caught before it ships twice.

When parallel agents make work cheap, overlap gets expensive. Alan compares work in progress across sessions and flags when two efforts are heading to the same place, so you can join forces instead of racing.

These sessions look like the same work

Same files, same goal, started a day apart

Platform · Maya, DevAdd rate limiting to the public APIrateLimit.ts · gateway.ts
Partner API · PriyaThrottle partner webhook callsthrottle.ts · gateway.ts

Both sessions are building a per-key limiter in the gateway. One shared implementation would cover both.

better. faster. TOGETHER.

Team from $10 per seat · Enterprise on contract