Skip to content

Add read-only raw access#3

Closed
davidlsneider wants to merge 47 commits intomainfrom
codex/raw-readonly
Closed

Add read-only raw access#3
davidlsneider wants to merge 47 commits intomainfrom
codex/raw-readonly

Conversation

@davidlsneider
Copy link
Owner

Summary

  • add read-only raw endpoints for authenticated pages scoped to union of API-key secrets
  • add read-only token model/service plus mint/list/revoke endpoints
  • enforce raw host isolation, GET-only access, rate limiting, audit logging, and response metadata
  • document requirements and add E2E coverage for raw access

Testing

  • not run (migration required)

Notes

  • Prisma migration required for ReadOnlyToken + ReadOnlyTokenSecret (not included in this PR)
  • Raw host isolation uses RAW_DOMAIN if set, otherwise hostnames starting with raw.

davidlsneider and others added 30 commits February 11, 2026 15:25
* Fix skill install links to point to agent-skills repo

The "other agents" install commands were pointing to
HeyVincent-ai/Vincent/{skill} but the skills are published
to the agent-skills repo. Updated to use
HeyVincent-ai/agent-skills/{skill} instead.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Update Discord invite link

---------

Co-authored-by: David Sneider <davidsneider@Davids-MacBook-Pro-2.local>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
The "other agents" install commands were pointing to
HeyVincent-ai/Vincent/{skill} but the skills are published
to the agent-skills repo. Updated to use
HeyVincent-ai/agent-skills/{skill} instead.

Co-authored-by: David Sneider <davidsneider@Davids-MacBook-Pro-2.local>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Replaces placeholder Lit Protocol terms with Vincent-specific TOS structured
as a cloud provider rental agreement. Key material sections modeled after
Turnkey's non-custodial approach (encrypted keys on our infra, no raw key
access, Policy Rules, Key Export/Import). Adds AI agent risk disclosures,
rental fee structure, and service tier provisions.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Chris Axolotl Bot <chris@litprotocol.com>
* Add scheduled Sentry triage automation and summaries

* Switch Sentry automation from GitHub Actions to local cron

* Load local env file for cron triage runner

---------

Co-authored-by: Chris Axolotl Bot <chris@litprotocol.com>
* fix tests, add CI

* fix ci

* lint fixes

* eslint config

* fix test file linting
* take ownership plan

* self custody ready for testing

* skill update

* published

* take ownership improvements

* walletconnect, fix typo

* walletconnect almost working

* yay self custody for evm wallets works

* fix address passing

* self custody actually works

* frontend env cleanup

* fix race condition

* Remove if statement if zerodev env var not provided

* ternary expression fix

* Take ownership stores challenge in db

* update docs

* sig validation

* update plans

* Test written

* fix issue with legacy wallets

* min for zerodev api key

* Update src/services/ownership.service.ts

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

* eslint fix

* npm i

---------

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* Add phase 3 Sentry noise filtering for actionable alerts

* Add draft fix-attempt PR automation for safe Sentry candidates

---------

Co-authored-by: Chris Axolotl Bot <chris@litprotocol.com>
Co-authored-by: Chris Cassano <1285652+glitch003@users.noreply.github.com>
* track chains transferred

* Remove unrelated CSP changes from this PR

The CSP updates were unrelated to chains-transferred tracking
and were accidentally included. They belong in a separate PR.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* basic plan

* new plan

* phases 1 through 3 mostly done

* datasources and preinstall

* empty commit

* data source credit fix

* fix ci

* add skills
* speed up railway build

* trying dockerfile

* try again

* try again

* vite env vars

* non root fix
* api docs

* copilot fixes
* Initial plan

* Add POLYMARKET_WALLET option to create secret dropdown

Co-authored-by: glitch003 <1285652+glitch003@users.noreply.github.com>

---------

Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: glitch003 <1285652+glitch003@users.noreply.github.com>
glitch003 and others added 17 commits February 15, 2026 13:00
* better docs

* docs link
* some phases completed

* Add all skill tests and fix agent harness for AI SDK v6

- Fix model ID (google/gemini-2.5-flash, not -preview)
- Fix body double-serialization in http_request tool
- Use AI SDK v6 API (stopWhen/stepCountIs, inputSchema, staticToolCalls)
- Add brave-search, twitter, polymarket tests
- All 4 tests pass against production (~15s total)

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* faster builds

* another fix

* fix

* coilot fix
* Add Stytch auth flow for data source tests

Brave-search and twitter tests now create a properly claimed
DATA_SOURCES secret via Stytch sandbox magic link auth before
running the agent, so they verify real search results (200 + data)
instead of accepting 403/402 errors.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Address copilot review feedback on skill CI tests

- Add explicit null check for claim URL token instead of non-null assertion
- Document the Stytch sandbox magic link token with reference link
- Add afterAll cleanup to delete test secrets in both twitter and brave-search tests
- Export deleteSecret helper from auth module

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* Add Stytch auth flow for data source tests

Brave-search and twitter tests now create a properly claimed
DATA_SOURCES secret via Stytch sandbox magic link auth before
running the agent, so they verify real search results (200 + data)
instead of accepting 403/402 errors.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Address copilot review feedback on skill CI tests

- Add explicit null check for claim URL token instead of non-null assertion
- Document the Stytch sandbox magic link token with reference link
- Add afterAll cleanup to delete test secrets in both twitter and brave-search tests
- Export deleteSecret helper from auth module

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* script to run manually

---------

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* claude autoreview

* comment fix

* address feedback
@davidlsneider
Copy link
Owner Author

Closing in favor of upstream PR: https://github.com/HeyVincent-ai/Vincent/pull/91

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants