chore: version packages#3001
Open
gram-bot[bot] wants to merge 1 commit into
Open
Conversation
There was a problem hiding this comment.
Claude Code Review
This repository is configured for manual code reviews. Comment @claude review to trigger a review and subscribe this PR to future pushes, or @claude review once for a one-time review.
Tip: disable this comment in your organization's Code Review settings.
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
Collaborator
🚀 Preview Environment (PR #3001)Preview URL: https://pr-3001.dev.getgram.ai
Gram Preview Bot |
5100a53 to
8d4a444
Compare
8d4a444 to
2ebfd30
Compare
2ebfd30 to
914dfd8
Compare
914dfd8 to
bd981e3
Compare
bd981e3 to
49d2d00
Compare
49d2d00 to
393007d
Compare
393007d to
87818f8
Compare
87818f8 to
8419d57
Compare
8419d57 to
ed7ef3b
Compare
ed7ef3b to
9de93b2
Compare
9de93b2 to
ebd242a
Compare
ebd242a to
6fa92dd
Compare
6fa92dd to
b2fedab
Compare
b2fedab to
45fdcfa
Compare
Releases: server@0.59.0 dashboard@0.60.0
45fdcfa to
a7410f0
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This PR was opened by the Changesets release GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated.
Releases
dashboard@0.60.0
Minor Changes
b58bf0f: Adds an org-level AI Integrations product surface with Cursor as the first provider. Organization admins can connect a Cursor Admin API key from org settings, and an hourly Temporal workflow polls Cursor for token and cost usage events and writes them into ClickHouse
telemetry_logsso the dashboard shows Cursor usage and cost alongside Claude Code data. The dashboard cost copy is updated to reflect Cursor and Claude Code coverage, and the employee detail page now shows cost beside total tokens.ed12a35: Add multiple role support to the RBAC system. Users can now be assigned multiple roles simultaneously, replacing the previous single-role assignment model.
3b8bfb4: Adds
risk.results.listForAgent— a redacted variant ofrisk.results.listfor AI assistant / MCP consumption. The new endpoint returns the same fields aslistRiskResultsbut replaces thematchfield withmatch_redacted, an opaque token of the form<redacted len=N sha=XXXXXXXX>whereNis the byte length andXXXXXXXXis the first 8 hex characters ofsha256(match). Identical secrets produce identical fingerprints so agents can dedupe leak counts without ever seeing secret content.shadow_mcpfindings passmatchthrough verbatim because the value is a server URL or stdio command identifier (already shown unmasked in the dashboard), and exact byte positions are coarsened to a singleposition_knownboolean to remove reconstruction signals.The dashboard's AI Insights sidebar gains risk-aware suggestions on the Security Overview and Policy Center pages, plus a system-prompt rule that bars the assistant from echoing
match_redactedvalues verbatim.Patch Changes
remote_mcp_server_idfilter.TelemetrySummaryRowandToolBarListare extracted into a sharedSourceActivityPanelcomponent consumed by both the OpenAPI and Remote MCP source overview tabs.server@0.59.0
Minor Changes
5f4c259: Add admin API endpoints for managing organizations and OAuth/OIDC configuration, protected by a dedicated admin security middleware. Includes a mock OIDC server for local development and testing.
0c431a0: initial MCP resource method interceptors
8e247f9: Chat loading is now paginated by generation, returning one generation per request. The chat detail panel fetches older generations in parallel until the full transcript is assembled, so long-running sessions no longer stall on the initial fetch.
b58bf0f: Adds an org-level AI Integrations product surface with Cursor as the first provider. Organization admins can connect a Cursor Admin API key from org settings, and an hourly Temporal workflow polls Cursor for token and cost usage events and writes them into ClickHouse
telemetry_logsso the dashboard shows Cursor usage and cost alongside Claude Code data. The dashboard cost copy is updated to reflect Cursor and Claude Code coverage, and the employee detail page now shows cost beside total tokens.ed12a35: Add multiple role support to the RBAC system. Users can now be assigned multiple roles simultaneously, replacing the previous single-role assignment model.
3b8bfb4: Adds
risk.results.listForAgent— a redacted variant ofrisk.results.listfor AI assistant / MCP consumption. The new endpoint returns the same fields aslistRiskResultsbut replaces thematchfield withmatch_redacted, an opaque token of the form<redacted len=N sha=XXXXXXXX>whereNis the byte length andXXXXXXXXis the first 8 hex characters ofsha256(match). Identical secrets produce identical fingerprints so agents can dedupe leak counts without ever seeing secret content.shadow_mcpfindings passmatchthrough verbatim because the value is a server URL or stdio command identifier (already shown unmasked in the dashboard), and exact byte positions are coarsened to a singleposition_knownboolean to remove reconstruction signals.The dashboard's AI Insights sidebar gains risk-aware suggestions on the Security Overview and Policy Center pages, plus a system-prompt rule that bars the assistant from echoing
match_redactedvalues verbatim.Patch Changes
9d6ba7b:
/rpc/telemetry.getObservabilityOverviewnow accepts an optionalremote_mcp_server_idfilter so callers can scope summary, time-series, and per-tool breakdown metrics to a single Remote MCP source. Combinable with the existingtoolset_slugfilter.9d6ba7b:
/x/mcptools/call traffic now writes a structured row to ClickHousetelemetry_logsper invocation, mirroring the existing/mcpemit. The row carriesgram.remote_mcp_server.idandgram.tool.nameattributes so the Source Activity panel for a Remote MCP source can filter telemetry by the originating remote server. Emission is fire-and-forget so ClickHouse latency does not appear in tool-call tail latency.fae81e1: Public-MCP
/authorizeaccepts a newrequireUserIdentity=1query parameter that forces the caller through the IDP so the resulting session is bound to a user subject rather than an anonymous one. Without the parameter, public-toolset/authorizecontinues to mint an anonymous subject regardless of ambient cookies or Bearer tokens. Callers from outside the endpoint's organization receive a 403 from the IDP callback — public toolsets that need cross-organization access should omit the parameter and use anonymous sessions.The assistant runtime sets the parameter when initiating MCP authorization flows against Gram-served endpoints so subsequent tool calls can be attributed to the user. Foreign (non-Gram) authorization endpoints discovered via
.well-known/oauth-authorization-serverdo not receive the parameter.d4ab97a: Assistants are now instructed to treat OAuth/MCP authentication as owner-only and to avoid pre-emptively prompting for auth on toolsets they have not yet needed.
508aef1: Always emit the
resultfield in JSON-RPC success responses from the MCP server. Empty-result handlers (notablyping) previously sent{"jsonrpc":"2.0","id":N}, which violates JSON-RPC 2.0 and the MCP spec. Cursor's MCP SDK rejected those frames withinvalid_unionzod errors and dropped the transport to a failed state after each keep-alive ping.