Individual developers.
Move from idea to verified diff without procurement.
For developers
Build without asking the meter.
Use a serious coding agent without asking finance, watching a token meter, or rationing your best ideas.
Migrate auth module to argon2id, tests green.running
Cost
$1.84/ $5.00
Steps
287/ 1000
Wall time
3h 22m/ 24h
Completion
62%2 / 3 criteria
iter 48 · re-plan in 12 steps
now
Refactoring src/auth/session.ts to call the new hashing API.
▾str_replacesrc/auth/session.ts
88−const hash = await bcrypt.hash(password, 12);88+const hash = await argon2.hash(password);
bashpnpm test src/auth/session.test.ts
RUN v1.6.0 /api
✓ verify · accepts current session (8 ms)
✓ verify · rejects expired session (3 ms)
✓ hash · argon2id round-trip (112 ms)
· running migration compat suite
✓ verify · accepts current session (8 ms)
✓ verify · rejects expired session (3 ms)
✓ hash · argon2id round-trip (112 ms)
· running migration compat suite
memory writes · this run
decision
Bridge bcrypt verifies for 90 days, then auto-migrate on next login.
fact
Auth reference cases hold pre-migration hashes.
decision
Drop bcrypt dependency after final migration window.
style
Use argon2.verify over manual hash compare. Reinforced.
◇ SponsoredRuntime patron—Supporting managed model access for builders.·localhost:3000/sponsorsLearn why this appears. Dismiss.
∠goal/argon2idtools287ctx184K / 200Kcost$1.84modegoaliter 48 · acting
Stay in the terminal.
Open a repo, describe the task, and let Picasso gather context. It can plan, edit, run checks, debug failures, and explain the result without pulling you into a new workspace.
developer flowterminal to Studio
repoopen canvas
sketchreview plan
checktests pass
replayinspect later
Students.
Learn from visible plans, commands, failures, and explanations.
Freelancers.
Carry repo memory across client work without starting cold.
Small teams.
Use the same agent harness without seat pricing in the way.
Bring any model path.
Use the managed path, a supported Claude or ChatGPT subscription, provider API keys, or local models through Ollama and vLLM.
model accessfour paths
Managedno setupsponsor-supported
Claude / ChatGPTsubscriptionpassthrough
Provider keysBYOdirect billing
LocalOllama / vLLMyour machine
Managed.
Picasso pays for inference through sponsor-supported access.
Subscription.
Use an existing provider subscription where supported.
API key.
Keep billing in your provider account.
Local.
Use your hardware and your models.
Local when privacy or offline work matters.
For privacy-sensitive or offline experiments, use local mode with Ollama or vLLM. Hosted providers, managed inference, cloud sync, sponsor surfaces, and remote MCP servers still require network access.
Private experiments.
Keep inference on hardware you control when the task does not need hosted models.
Explicit network use.
Hosted routes, sync, sponsors, and remote tools stay tied to network-dependent modes.
Same loop.
Plans, goals, memory, replay, and subagents keep the same shape across local and hosted routes.
Let Picasso remember the repo.
Memory keeps architecture, commands, style preferences, and decisions available when they matter. It is retrieval only, visible, editable, exportable, and deletable.
Project facts.
How the repo is shaped.
Style.
How you like code reviewed and tested.
Decisions.
Why a path was chosen.
Run bigger tasks with `/goal`.
Ask Picasso to remove a deprecated API, repair a test suite, migrate a module, fix a build, or complete a feature against clear criteria.
agent loopplan / act / observe / reflect
01Planread repo, draft plan
02Actedit files, run tools
03Observetests, logs, diffs
04Reflectcontinue or stop
completion criteria
Define done.
Set completion criteria up front.
Keep state.
Continue across sessions.
Inspect afterward.
Replay the canvas in the Studio.
Free does not mean careless.
The managed path is funded by sponsors, not by turning your code into inventory. Sponsor content is labeled and kept out of code, model output, tool results, plans, routing, memory, and subagents. Budget checks can stop by provider, user, org, session, goal, or sponsored unlock instead of silently continuing.
trust boundaryoutside the agent loop
generated codeclean
model outputclean
tool resultsclean
plan documentsclean
routingnot influenced
memoryretrieval only