agentfootprint

Agents that explain themselves.

01

BUILD

Snap an agent together from pieces.

what's inside
LLMCallAgentSequenceParallelConditionalLoopSwarmpatterns
Read more →
import { anthropic } from 'agentfootprint/llm-providers'
const scout = Agent.create({
provider: anthropic({ defaultModel: 'claude-sonnet-4-5-20250929' }),
model: 'anthropic',
})
.system('cite every claim')
.tool(webSearch)
.tool(readPdf)
.build()
await scout.run({ message: 'summarise q3-brief.pdf' })
02

TEACH

Give it skills — it opens only the one it needs.

what's inside
skillssteeringfactsinjectionsskill graphtriggers
Read more →
read('q3-brief.pdf')
web-search
read-pdf
summarize
cite-check
03

REMEMBER

It keeps what matters between conversations.

what's inside
episodicsemanticnarrativeRAGRedisin-memory
Read more →
04

SWAP

Same agent. Claude, GPT, or your laptop.

what's inside
anthropicopenaiazurebedrockollamabrowsermockbring-your-own
Read more →
import { anthropic } from 'agentfootprint/llm-providers'
const scout = Agent.create({
provider: anthropic({ defaultModel: 'claude-sonnet-4-5-20250929' }),
model: 'anthropic',
})
.system('cite every claim')
.tool(webSearch)
.build()
05

WATCH

See it think, in real time.

what's inside
67 eventsflowcharttimelinecostlive statusnarrative
Read more →
live · evt 67
plansearchreadverifyanswer
66 tool_call web.fetch · 212ms · $0.0031
67 llm_call verify ·
06

ASK

It checks in before doing anything big — with evidence.

what's inside
checkInpause/resumepermissionsaudit trail
Read more →
check-inrun #418 · paused
WILL DOrm ./cache — 3 files
READconfig.ts · .env.example
WHAT DROVE THISplan step 4 · rule: fs.delete → confirm
07

WHY

When something's wrong, ask why. It answers with receipts.

what's inside
localizeablation proofwalk to rootinfluenceself-explain
Read more →
$4.2M
SUSPECT CONTEXT · RANKED BY INFLUENCE
1doc: q3-brief.pdf §2.82PROVEN
2mem: pricing-2024.41
3tool: web.search #3.17
4sys: persona.09
08

SURVIVE

Outages happen. Runs continue.

what's inside
retryfallbackcircuit breakerreliability rulescheckpoints
Read more →
run #418anthropic openai ← rerouted
ckpt 12resumed · step 12
npm install agentfootprint