chore(acp): bump glm-acp-agent, grok and nova registry pins to probed versions - #865
Merged
Merged
Conversation
kaizhou-lab
added a commit
that referenced
this pull request
Aug 17, 2026
## Summary Scheduled ACP Registry version sync. One npx pin drifted since #865, backed by a fresh ACP probe of the exact pinned version. Lock-only change — one line; no metadata, migration, or entrypoint edits, and no lock-derived test assertion embeds this version. | backend | package | old → new | initialize | session/new | |---|---|---|---|---| | dimcode | `dimcode` | 0.3.13 → **0.3.15** | ok (protocolVersion 1, agentInfo version 0.3.15) | auth required (`-32000`, "Provider credentials are required") | Meets the release-lock criterion: `initialize` succeeds and `session/new` returns a clearly classified authentication requirement. The probe ran with no inherited HOME or credentials. The Registry advertises 0.3.15 directly (0.3.14 is skipped), and 0.3.15 is what was probed and pinned. The other 10 Registry-pinned packages match the snapshot exactly — no drift. In particular glm-acp-agent 1.5.0, grok 1.0.5 and nova 1.1.35, all bumped last night in #865, are still current. Package names and entrypoint args are unchanged for all 11. Drifted but not upgraded: none. `mimo-code` remains the one non-Registry builtin (no `registry_json_id`), excluded from drift reconciliation. dimcode continues to self-report `agentInfo.title` as "DimAgent" while the public Registry card and CDN entry still say "DimCode" (`dimcode.dev` unchanged), so the seeded display name is untouched — the same standing observation carried since #814. ## Registry snapshot - Audit pinned to release tag [`v2026.08.17-168e146`](https://cdn.agentclientprotocol.com/registry/v1/v2026.08.17-168e146/registry.json) of `agentclientprotocol/registry`, fetched via the versioned CDN path for reproducibility. - No newly listed and no delisted agents versus the previous baseline (38 ids). ## Validation - `just migration-check` — pass - `just lint-fix` (`cargo fix` + `clippy --fix --workspace -D warnings`) — clean - `just fmt` — clean - **Local `cargo nextest` intentionally skipped, by standing policy for lock-only bumps** (established 2026-08-11; #810, #814, #822, #837, #848, #859 and #865 all merged green under it). The Test check on this PR is the authority for this change: the merge decision depends on CI rather than the local run, and this host's load only manufactures timeout-shaped test failures, which nothing in the local steps above is subject to. ## Logging No logging changes: lock-only version bump; existing startup/session error paths already identify a failing agent by backend. Co-authored-by: zk <>
kaizhou-lab
added a commit
that referenced
this pull request
Aug 19, 2026
… to probed versions (#880) ## Summary Scheduled ACP Registry version sync. Four npx pins drifted since #873, each backed by a fresh serial ACP probe of the exact pinned version. Lock-only change — four lines; no metadata, migration, or entrypoint edits, and no lock-derived test assertion embeds any of these versions. | backend | package | old → new | initialize | session/new | |---|---|---|---|---| | dimcode | `dimcode` | 0.3.15 → **0.3.16** | ok (agentInfo version 0.3.16) | auth required (`-32000`, "Provider credentials are required") | | dirac | `dirac-cli` | 0.4.36 → **0.4.37** | ok (agentInfo dirac 0.4.37) | success (modes plan/act) | | glm-acp-agent | `glm-acp-agent` | 1.5.0 → **1.6.0** | ok (protocolVersion 1) | success, unauthenticated — modes `default` / `accept_edits` / `bypass_permissions` | | grok | `@xai-official/grok` | 1.0.5 → **1.0.6** | ok (protocolVersion 1) | auth required (`-32000`, "no auth method id provided") | All four meet the release-lock criterion: `initialize` succeeds and `session/new` either succeeds or returns a clearly classified authentication requirement. Probes ran serially with no inherited HOME or credentials. **glm-acp-agent crosses another minor version, so its stored `yolo_id` was re-verified rather than assumed** — the same check #865 applied at 1.5.0. The 1.6.0 session catalog still advertises `bypass_permissions`, exactly the value migration 025 seeded, so the existing metadata stays correct and this remains a lock-only change. Had that mode disappeared, the `yolo_id` would need its own migration and would not belong in this PR. The other 7 Registry-pinned packages (autohand, codebuddy, deepagents, kilo, nova, pi, sigit) match the snapshot exactly. Package names and entrypoint args are unchanged for all 11. Drifted but not upgraded: none. `mimo-code` remains the one non-Registry builtin (no `registry_json_id`), excluded from drift reconciliation. Two long-running vendor self-report quirks persist, neither actionable: dimcode reports `agentInfo.title` as "DimAgent" against a public listing that says "DimCode" (`dimcode.dev` unchanged), and glm-acp-agent 1.6.0 reports `agentInfo.version` as `1.0.0`. The Registry card, package identity and entrypoint remain the authorities, and `initialize` succeeded in both cases. ## Registry snapshot - Audit pinned to release tag [`v2026.08.19-28cd4e9`](https://cdn.agentclientprotocol.com/registry/v1/v2026.08.19-28cd4e9/registry.json) of `agentclientprotocol/registry`, fetched via the versioned CDN path for reproducibility. - No newly listed and no delisted agents versus the previous baseline (38 ids). ## Validation - `just migration-check` — pass - `just lint-fix` (`cargo fix` + `clippy --fix --workspace -D warnings`) — clean - `just fmt` — clean - **Local `cargo nextest` intentionally skipped, by standing policy for lock-only bumps** (established 2026-08-11; #810, #814, #822, #837, #848, #859, #865 and #873 all merged green under it). The Test check on this PR is the authority for this change: the merge decision depends on CI rather than the local run, and this host's load only manufactures timeout-shaped test failures, which nothing in the local steps above is subject to. ## Logging No logging changes: lock-only version bumps; existing startup/session error paths already identify a failing agent by backend. Co-authored-by: zk <>
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
1 participant
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.
Summary
Scheduled ACP Registry version sync. Three npx pins drifted since #859, each backed by a fresh serial ACP probe of the exact pinned version. Lock-only change — three lines; no metadata, migration, or entrypoint edits, and no lock-derived test assertion embeds any of these versions.
glm-acp-agentz-ai-api-key)default/accept_edits/bypass_permissions@xai-official/grokgrok.com)-32000, "no auth method id provided")@compass-ai/novakore-terminal-auth)-32000, "Click Nova Setup to configure your API keys")All three meet the release-lock criterion:
initializesucceeds andsession/neweither succeeds or returns a clearly classified authentication/configuration requirement. Probes ran serially with no inherited HOME or credentials.glm-acp-agent crosses two minor versions (1.3.0 → 1.5.0), so its stored
yolo_idwas re-checked rather than assumed. The 1.5.0 session catalog still advertisesbypass_permissions, which is exactly the value migration 025 seeded, so the existing metadata remains correct and this stays a lock-only change. Had that mode disappeared, theyolo_idwould need its own migration and would not belong in this PR.The other 8 Registry-pinned packages (autohand, codebuddy, deepagents, dimcode, dirac, kilo, pi, sigit) match the snapshot exactly. Package names and entrypoint args are unchanged for all 11. Drifted but not upgraded: none.
mimo-coderemains the one non-Registry builtin (noregistry_json_id), excluded from drift reconciliation.Vendor self-reported version mismatches — three of them now, none blocking: glm-acp-agent 1.5.0 reports
agentInfo.versionas1.0.0; nova 1.1.35 reportskore-cli1.0.0; dimcode reportsagentInfo.title"DimAgent" against a public listing that says "DimCode". In every case the Registry card, the package identity and the entrypoint are the authorities, andinitializesucceeded — the pins are evidenced regardless of what the handshake calls itself.Registry snapshot
v2026.08.16-e3c983dofagentclientprotocol/registry, fetched via the versioned CDN path for reproducibility.Validation
just migration-check— passjust lint-fix(cargo fix+clippy --fix --workspace -D warnings) — cleanjust fmt— cleancargo nextestintentionally skipped, by standing policy for lock-only bumps (established 2026-08-11; chore(acp): bump dimcode, dirac, grok and kilo registry pins to probed versions #810, chore(acp): bump dimcode registry pin to probed 0.3.9 #814, chore(acp): bump dimcode, grok, kilo and sigit registry pins to probed versions #822, chore(acp): bump dimcode, dirac, grok and nova registry pins to probed versions #837, chore(acp): bump dimcode and nova registry pins to probed versions #848 and chore(acp): bump dimcode, dirac, grok and kilo registry pins to probed versions #859 all merged green under it). The Test check on this PR is the authority for this change: the merge decision depends on CI rather than the local run, and this host's load only manufactures timeout-shaped test failures, which nothing in the local steps above is subject to.Logging
No logging changes: lock-only version bumps; existing startup/session error paths already identify a failing agent by backend.