Skip to content

[deps-release-notes] copilot-cli — upstream release action items #2041

Closed as not planned

Description

@github-actions

Rolling upstream release action items — copilot-cli

This is the single canonical tracking issue for action items arising from
new releases of the copilot-cli dependency. The update-awf-version workflow
appends a new comment to this issue for each version bump going forward, so
the most recent activity lives in the comments below. This body is a
consolidated history of everything filed so far.

Latest pinned version covered: 1.0.81

Consolidated history (earliest → latest)

1.0.48 → 1.0.56 (was #803)

  • Security: secret scanning covers commit messages + PR descriptions (v1.0.51): Redacts secrets before publishing.
  • Security: permissions.disableBypassPermissionsMode (v1.0.55): Prevents enabling allow-all/yolo mode; relevant for sandboxed pipelines.
  • --session-id=<id> flag (v1.0.51): Resume/start sessions with a specific UUID — useful for pipeline correlation.
  • postToolUse hook additionalContext injected as system message (v1.0.49): Instead of being discarded.
  • preMcpToolCall hook (v1.0.51): Control outgoing MCP request metadata.
  • deferred-tool-loading frontmatter for custom agents (v1.0.52): Reduces initial token overhead.
  • MCP tools with both content + structuredContent surface both (v1.0.56): Review ado-aw MCP tools to ensure dual output is intentional.
  • Context window tier selection (v1.0.52): Default ~200K vs 1M tokens enforced end-to-end.
  • Claude Opus 4.8 support (v1.0.55).
  • Deprecation: oauth.clientId/oauth.callbackPort → oauthClientId/auth.redirectPort (v1.0.52).

1.0.48 → 1.0.59 (was #843)

  • Breaking: preToolUse hook errors now deny the tool call (v1.0.57): Previously hook errors silently allowed execution; now they block the call.
  • Breaking: Remote JSON RPC enabled by default (v1.0.58): Verify this doesn't open unexpected network surface in AWF-sandboxed environments.
  • Breaking: --plugin-dir skills take precedence over personal-home skills (v1.0.55): Skill resolution order: project > plugin-dir > personal > custom.
  • ADO-only repos: built-in GitHub MCP server exposes only web_search (v1.0.57): Directly relevant to ado-aw; consider whether the GitHub MCP extension should account for ADO-repository detection.

1.0.48 → 1.0.60 (was #881)

  • Alpine Linux (musl libc) support (v1.0.49): Now runs on Alpine-based containers.
  • Rubber Duck agent enabled by default (v1.0.58): Check whether this adds unexpected turn overhead in automated runs.

1.0.60 → 1.0.61 (was #951)

  • Auto-load MCP servers from .github/mcp.json (v1.0.61): ado-aw uses --additional-mcp-config; consider guarding against auto-loading interference.
  • Claude Fable 5 model support (v1.0.61): Add to validated models in src/engine.rs/docs/engine.md.
  • mTLS + private-CA for OTLP telemetry (v1.0.61): Added http/protobuf OTLP export via OTel env vars.
  • Bug fix: grep/glob single-path fix (v1.0.61): Previously missed results with single path argument.
  • Bug fix: shell validation false positives (v1.0.61): Fixed blocking of harmless commands with words like "kill" in string literals/heredocs.

1.0.60 → 1.0.62 (was #1002)

  • Breaking: write_bash interactive input removed (v1.0.62): Shell commands now use lightweight process spawning; interactive input no longer supported. ado-aw exposes write_bash as a first-class tool — assess whether tool surface/docs need updating.
  • Configure subagent model, reasoning effort, context tier (v1.0.62): Consider surfacing reasoning_effort/context-tier as front-matter fields.
  • Kerberos/SPNEGO proxy auth (v1.0.62): Auto-auth through corporate forward proxies; relevant for enterprise ADO environments.

1.0.62 → 1.0.63 (was #1057)

  • deferTools MCP server config option (v1.0.63): Keeps server tools available even when tool search is enabled. ado-aw could use this for safe-outputs and GitHub MCP servers to prevent them from being filtered out.

1.0.62 → 1.0.64 (was #1183)

  • New OTel cache + reasoning token fields (v1.0.64): gen_ai.usage.cache_read.input_tokens, gen_ai.usage.cache_creation.input_tokens, gen_ai.usage.reasoning.output_tokens per GenAI semconv (old underscore-separated names gone). Update src/agent_stats.rs to parse new attribute names.
  • Static OAuth client overrides for MCP server auth (v1.0.64): ado-aw could expose in MCP server config front-matter.
  • Autopilot mode auto-handles ask_user, elicitation, sampling, permission prompts (v1.0.64): Reliability improvement for unattended pipeline runs.
  • Deprecation: report_intent tool removed (v1.0.64): INTERNAL_TOOL_NAMES exclusion in src/agent_stats.rs containing "execute_tool report_intent" is now dead code; can be cleaned up.

1.0.64 → 1.0.67 (was #1272)

  • Claude Sonnet 5 model support (v1.0.67): Add to src/engine.rs model allowlist + docs.
  • Subagent sessions inherit parent tool restrictions (v1.0.67): Security-relevant behavior change; verify against expected security model.
  • MCP OAuth against Microsoft Entra vanity domain servers (v1.0.67): Fixes AADSTS auth failures for Entra vanity domains; relevant for enterprise Azure AD MCP tools.
  • copilot skill subcommand (v1.0.65): New /skill alias for listing/adding/removing skills; may be relevant for agency plugin scaffolding.
  • Deprecation: Claude Opus 4.6 Fast → Claude Opus 4.8 Fast (v1.0.66): Update any ado-aw references to claude-opus-4.6-fast.

1.0.64 → 1.0.68 (was #1305)

  • Breaking: session limits require minimum 30 AI credits (v1.0.67): Pipelines with session-limit: below 30 AI credits will be rejected at runtime. Review compiler defaults.
  • Claude Opus 4.8 Fast model support (v1.0.66): Add claude-opus-4.8-fast to src/engine.rs allowlist.
  • kimi-k2.7-code model support (v1.0.68): Add to model allowlist.

1.0.70 → 1.0.81 (was #2041) — canonical

  • Breaking (v1.0.79): Sandbox settings keys renamed from sandbox.gitAuth/sandbox.ghAuth to sandbox.auth.git/sandbox.auth.gh, with no migration path. Old keys are silently ignored, and SDK requests using the old keys are rejected as invalid. Any ado-aw-generated engine config or docs referencing the old sandbox setting names need to be updated. See the v1.0.79 release.
  • Breaking (v1.0.79): --worktree default behavior changed — worktreeBaseRef now defaults to HEAD instead of the remote default branch. Could change output for any ado-aw workflow relying on worktree creation defaults. See the v1.0.79 release.
  • Security (v1.0.78): The /allow-all auto safety-judge model is no longer user-configurable — hardens against downgrade to a weaker safety judge. See the v1.0.78 release.
  • Security (v1.0.78): Managed-settings fetch failure behavior changed from fail-closed to fail-open — worth confirming this doesn't weaken ado-aw's sandboxed-agent posture when managed settings can't be reached. See the v1.0.78 release.
  • Security (v1.0.72): Sandbox macOS keychain access now defaults to off, reducing default credential exposure inside the sandboxed agent. See the v1.0.72 release.
  • v1.0.81: Support for MCP protocol version 2026-07-28, OpenTelemetry trace-context propagation (traceparent/tracestate) into hooks, and per-agent usage metrics via --usage-output-file — could improve ado-aw's OTel-based agent stats correlation (src/agent_stats.rs). See the v1.0.81 release.
  • v1.0.81: Windows Entra ID broker-based MCP auth (no-prompt sign-in) — relevant if ado-aw ever needs Windows-hosted agent auth flows.
  • v1.0.81: Plan mode now hard-blocks mutating tool calls — a safety property ado-aw could rely on/document for plan-only agent configurations.
  • v1.0.76/v1.0.77: Enterprise admins can enforce a restrictive sandbox floor and managed sandbox policy via macOS/Windows native MDM — relevant to ado-aw's OneBranch sandboxing story for enterprise consumers. See the v1.0.76 release and v1.0.77 release.
  • v1.0.72: Opt-in git/gh auth inside the OS sandbox — worth evaluating against ado-aw's own credential-isolation model (ado-proxy).

Consolidated by the Deps Release-Notes Consolidator workflow. Superseded per-release issues were closed and point here.> Generated by Deps Release-Notes Consolidator · auto · 76.3 AIC · ⌖ 11.6 AIC · ⊞ 10.6K · ◷

Activity

  1. github-actions commented on Sep 5, 2026

    @github-actions
    ContributorAuthor

    copilot-cli 1.0.70 → 1.0.83

    Releases analyzed

    Breaking changes

    • Plan mode now hard-blocks built-in tool calls that would modify the workspace (v1.0.71): the agent can no longer edit files or run mutating shell commands while planning; MCP and external tools are still allowed. Relevant if any ado-aw agent relies on plan-mode file mutation.
    • sandbox.gitAuth/sandbox.ghAuth settings keys renamed to sandbox.auth.git/sandbox.auth.gh, no migration (v1.0.79): old keys are silently ignored in settings files, and SDK requests sending the old keys are now rejected as invalid rather than ignored.
    • --worktree now defaults to starting from HEAD instead of the remote default branch (v1.0.79), controllable via the new worktreeBaseRef setting.
    • On macOS and Linux, sandboxed commands can no longer reach services running on the host machine, including a server the sandboxed command itself starts on 127.0.0.1 (v1.0.81): test suites that bind a local port will fail unless "Allow local network" is enabled in /sandbox. Worth checking whether any ado-aw sandboxed agent workflow binds local ports.
    • Linux sandboxing now requires slirp4netns, nsenter, iptables, ip6tables, iptables-restore and ip6tables-restore on PATH (v1.0.81/v1.0.83): sandboxed commands will start failing to launch without these; relevant to the AWF/OneBranch sandbox image if copilot's own sandbox mode is ever exercised inside it.
    • Removed the PLUGINS_DASHBOARD opt-out and the legacy skills picker (v1.0.81): /skills, bare /mcp, and /mcp show always open the dashboard now; /plugins command removed (moved to /plugin, /mcp, /skills).

    Security fixes

    • permissions.disableBypassPermissionsMode unrecognized values now enforced as disable instead of rejected (v1.0.81): hardens enterprise-managed policy handling for the bypass/allow-all mode ado-aw's sandboxed agent must not enable.
    • forceRemoteSettingsRefresh now fails closed (v1.0.82 chain via v1.0.81 notes): cached managed-settings policy is never served as a fetch-failure fallback when this is set; until a fresh policy is fetched, non-default MCP servers are blocked and bypass-permissions mode cannot be enabled — a stricter default posture worth being aware of for network-isolated ado-aw runs.
    • An enterprise-denied MCP server can no longer start before the managed allow/deny policy resolves (v1.0.81): startup now waits for the managed-settings fetch instead of racing it.

    Notable features for ado-aw to adopt

    • --session-id-adjacent correlation and per-agent usage metrics in --usage-output-file JSON (v1.0.81): could improve OTel-based agent stats parsing (src/agent_stats.rs).
    • Hooks can receive the current OpenTelemetry trace context and emit correlated spans (traceparent/tracestate) (v1.0.81): potential integration point for ado-aw's audit/trace tooling.
    • models.list now includes service-published infoMessages/warningMessages per model (v1.0.81): could surface deprecation/availability warnings to ado-aw authors via catalog/lint.
    • --enable-mcp-server flag to re-enable MCP servers disabled in settings for a single run (v1.0.80): could be useful for MCPG diagnostics.
    • Add automatic HTTPS proxy mTLS client certificate support for model and web requests (v1.0.83): relevant to AWF's proxy-based network isolation model — worth checking compatibility.
    • Timeline headers show tool call duration and token usage exposed in ACP prompt results/live usage_update notifications (v1.0.78): more granular data ado-aw's audit analyzers could ingest.

    Deprecations

    • sandbox.gitAuth/sandbox.ghAuth → sandbox.auth.git/sandbox.auth.gh (v1.0.79) — see Breaking changes above; no migration path, old keys silently ignored.
    • /plugins command removed in favor of /plugin, /mcp, /skills (v1.0.81).

    Warning

    Firewall blocked 1 domain

    The following domain was blocked by the firewall during workflow execution:

    • index.crates.io

    To allow these domains, add them to the network.allowed list in your workflow frontmatter:

    network:
      allowed:
        - defaults
        - "index.crates.io"

    See Network Configuration for more information.

    Generated by Dependency Version Updater · auto · 116.7 AIC · ⌖ 10.7 AIC · ⊞ 16.5K · ◷

  2. github-actions commented on Sep 6, 2026

    @github-actions
    ContributorAuthor

    copilot-cli 1.0.81 → 1.0.83

    Releases analyzed

    Notable features for ado-aw to adopt

    • Add Client ID Metadata Document (CIMD) support for MCP OAuth sign-in (v1.0.83): New OAuth client-identification mechanism for MCP servers; relevant if ado-aw ever needs to register a CIMD for its own MCP tool servers instead of relying on ad-hoc client IDs.
    • Custom agents can list several models in model, tried in order until one is available, with model-policy: required (v1.0.83): A model-fallback list feature for custom agents. ado-aw's engine: front matter currently only supports a single model: value — this could be surfaced as a fallback list to improve reliability when a preferred model is temporarily unavailable.
    • Automatic HTTPS proxy mTLS client certificate support for model and web requests (v1.0.83): Relevant to ado-aw's AWF/api-proxy network path if a consumer's environment requires mTLS client certs for outbound model traffic.
    • Sandboxed file tools now read the same developer-tool paths as sandboxed shell commands, including token-bearing developer config such as ~/.npmrc; sandbox.allowDevToolAccess can disable this (v1.0.83): Worth double-checking against ado-aw's own credential-isolation assumptions in the AWF sandbox — files like ~/.npmrc can carry registry auth tokens, and this widens what sandboxed file-read tools can see by default.
    • Sandboxed gh commands now authenticate as the account configured for the repository instead of the Copilot CLI login (v1.0.83): Behavior change for any sandboxed gh invocation inside the agent; worth confirming this doesn't conflict with ado-aw's own GitHub MCP / credential-isolation model.

    No breaking changes, security-labeled fixes, or deprecations were called out in the v1.0.82/v1.0.83 release notes beyond the items already logged in the prior 1.0.81 comment on this issue.

    Warning

    Firewall blocked 1 domain

    The following domain was blocked by the firewall during workflow execution:

    • index.crates.io

    To allow these domains, add them to the network.allowed list in your workflow frontmatter:

    network:
      allowed:
        - defaults
        - "index.crates.io"

    See Network Configuration for more information.

    Generated by Dependency Version Updater · auto · 200.8 AIC · ⌖ 5.47 AIC · ⊞ 16.5K · ◷

  3. github-actions commented on Sep 12, 2026

    @github-actions
    ContributorAuthor

    copilot-cli 1.0.70 → 1.0.83

    Releases analyzed

    Breaking changes

    • sandbox.gitAuth/sandbox.ghAuth settings keys renamed to sandbox.auth.git/sandbox.auth.gh with no migration (v1.0.79) — old keys are silently ignored in settings files, and SDK requests that still send them are rejected as invalid. If ado-aw ever generates or documents settings.json/SDK-request payloads for the Copilot CLI engine (e.g. sandbox git/gh auth config), those references need updating to the new keys.
    • Linux sandboxing now requires slirp4netns, nsenter, iptables, ip6tables, iptables-restore, ip6tables-restore on PATH; proxy-mode egress restriction additionally requires util-linux 2.35+ and /dev/net/tun (v1.0.83) — the ado-aw agent sandbox image/runner (OneBranch build agents invoking the copilot CLI) must have these binaries available or sandboxed shell commands will fail to launch after upgrading to 1.0.83.
    • Sandboxed commands on macOS/Linux can no longer reach services on the host machine, including a server the command itself starts on 127.0.0.1 (v1.0.83) — any ado-aw agent workflow whose bash tool binds a local port (e.g. for a test server) will need Allow local network enabled in /sandbox, or an equivalent CLI/config flag if ado-aw wires sandbox settings.
    • Managed settings now fail open instead of fail-closed when a server-managed settings fetch fails and no usable cached policy exists (v1.0.78) — a behavior change for any enterprise-managed policy enforcement ado-aw relies on via COPILOT_CLI_VERSION-pinned engine config; worth confirming this doesn't weaken assumptions in docs/engine.md.

    Security fixes

    • copilot skill list strips terminal control characters from skill names/descriptions, preventing a crafted skill from injecting ANSI escape sequences into listing output (v1.0.72).
    • Enforce the sandbox filesystem policy on LSP file reads and rename edits (v1.0.71) — closes a gap where LSP-driven file access could bypass sandbox allowWrite/allowRead boundaries.
    • Sandbox denied paths enforced for relative and symlinked entries on macOS/Linux (v1.0.76) — closes a path-traversal-style bypass of sandbox deny rules.
    • Hook output bounded at 10 MiB per invocation (v1.0.76) — prevents an HTTP or command hook from exhausting memory via an unbounded response; also fixes several userPromptSubmitted hook edge cases that could corrupt a session.
    • Enterprise-denied MCP servers can no longer start before the managed allow/deny policy resolves (v1.0.83) — closes a startup race that could let a denied MCP server briefly run.
    • Sandboxed gh commands now authenticate as the account configured for the repository instead of the Copilot CLI login (v1.0.83) — reduces the risk of using the wrong credential scope for gh operations inside the sandbox.

    Notable features for ado-aw to adopt

    • Client ID Metadata Document (CIMD) support for MCP OAuth sign-in (v1.0.83) — relevant if ado-aw's azure-devops or other MCP tool integrations need standards-based OAuth client registration.
    • Automatic HTTPS proxy mTLS client certificate support for model and web requests (v1.0.83) — could simplify ado-aw's ado-proxy/API-proxy TLS story if the engine now handles client-cert auth natively.
    • Windows sign-in for remote MCP servers protected by Microsoft Entra ID via the OS authentication broker (WAM) (v1.0.81) — potential low-friction auth path for Windows-hosted MCP servers ado-aw wires up.
    • Add --with-token to copilot login to read an auth token from stdin (v1.0.81) — could simplify non-interactive COPILOT_CLI_VERSION-pinned login flows in the Agent stage compared to existing token-injection mechanisms.
    • Ship MCP 2026-07-28 protocol support to CLI, SDK, IDE, and in-memory clients (v1.0.81) — confirms wire-protocol compatibility for ado-aw's MCP tool surface at this pinned version.
    • Add per-agent usage metrics to --usage-output-file JSON output (v1.0.81) — could enrich src/agent_stats.rs's OTel-based token/usage parsing with a structured per-agent breakdown if ado-aw adopts this output format.

    Deprecations

    • None explicitly announced in this range. (/theme's deprecation-notice display logic was itself bug-fixed in v1.0.79, not a new deprecation.)

    Generated by Dependency Version Updater · auto · 195.5 AIC · ⌖ 3.17 AIC · ⊞ 14.8K · ◷

  4. github-actions commented on Sep 16, 2026

    @github-actions
    ContributorAuthor

    copilot-cli 1.0.70 → 1.0.85

    Releases analyzed

    Note: this comment covers only the increment beyond 1.0.81, since prior bumps up to that version were already recorded above.

    Breaking changes

    • Linux sandboxing now requires additional binaries on PATH (slirp4netns, nsenter, iptables, ip6tables, iptables-restore, ip6tables-restore) (v1.0.83) — sandboxed commands will fail to launch without these. Proxy-mode egress restriction additionally requires util-linux 2.35+ and /dev/net/tun access. Worth verifying these are present in the AWF/OneBranch agent image ado-aw compiles into, since ado-aw invokes copilot inside that sandbox.
    • macOS/Linux sandboxed commands can no longer reach services running on the host machine, including ones the command itself starts on 127.0.0.1 (v1.0.83) — this could break any ado-aw agent workflow that starts a local server and expects to curl localhost from the same sandboxed session; the release notes point at "Allow local network" in /sandbox as the opt-back-in.

    Notable features for ado-aw to adopt

    • Client ID Metadata Document (CIMD) support for MCP OAuth sign-in (v1.0.83) — could simplify OAuth-based MCP server configuration for user-defined stdio/HTTP MCP servers documented in docs/mcp.md.
    • copilot init now removes the .github directory it created when the run exits without writing an instructions file (v1.0.85) — behavior change relevant if ado-aw's own init flow or docs reference copilot init side effects.
    • Vim mode now available to everyone via /vim (v1.0.85) — purely an interactive-CLI UX feature, not applicable to ado-aw's non-interactive agent invocations.

    Deprecations

    • No new deprecations identified in this range (1.0.81 → 1.0.85) beyond UX/CLI surface changes already covered above; the remaining items were interactive-terminal UX fixes, model-picker/session-management polish, and internal refactors with no consumer-visible effect on ado-aw's non-interactive copilot invocation.

    Warning

    Firewall blocked 1 domain

    The following domain was blocked by the firewall during workflow execution:

    • index.crates.io

    To allow these domains, add them to the network.allowed list in your workflow frontmatter:

    network:
      allowed:
        - defaults
        - "index.crates.io"

    See Network Configuration for more information.

    Generated by Dependency Version Updater · auto · 278.5 AIC · ⌖ 5.29 AIC · ⊞ 14.8K · ◷

  5. github-actions commented on Sep 20, 2026

    @github-actions
    ContributorAuthor

    copilot-cli 1.0.70 → 1.0.86

    Releases analyzed

    Breaking changes

    • allowDevToolCaches sandbox setting renamed to allowDevToolAccess (v1.0.79): "BREAKING: the sandbox setting allowDevToolCaches is renamed allowDevToolAccess... The old key is no longer read and is ignored silently, so an existing false opt-out reverts to the default (on)." Any ado-aw-generated sandbox config or docs referencing allowDevToolCaches (see docs/network.md / sandbox-related front matter) must be updated to the new key, since a silent revert-to-default could widen dev-tool access unexpectedly for pipelines that had opted out.
    • /plugins command and PLUGINS_DASHBOARD opt-out removed (v1.0.81): "Removed /plugins; its resources moved to /plugin, /mcp and /skills... Removed the PLUGINS_DASHBOARD opt-out and the legacy skills picker it kept alive." Not directly agentic-workflow-facing (interactive-only surfaces), but any ado-aw docs/scripts referencing these commands should be checked.
    • Linux sandbox network/tooling requirements tightened (v1.0.83): "Linux sandboxing now needs slirp4netns, nsenter, iptables, ip6tables, iptables-restore and ip6tables-restore on PATH... Linux sandboxes now restrict network egress to the configured proxy; proxy mode requires slirp4netns, util-linux 2.35+, iptables, and /dev/net/tun access." If ado-aw's OneBranch/AWF sandbox images invoke copilot's own sandboxing (as opposed to relying solely on AWF's isolation), these new binary/kernel-capability requirements should be verified against the sandbox image.

    Security fixes

    • agentStop hook infinite-loop guard (v1.0.72): an agentStop hook that always blocks now force-ends the turn after 8 consecutive blocks, with a stop_hook_active flag for hook authors to detect forced continuation.
    • Bounded hook output (10 MiB) and stricter hook return-value validation (v1.0.76): guards against a malicious/misbehaving hook exhausting memory or corrupting session state via unbounded or malformed userPromptSubmitted hook responses.
    • Skill listing strips terminal control characters (v1.0.72): "a crafted skill can no longer inject ANSI escape sequences into the listing output" — relevant since ado-aw ships/consumes skills via agency/plugins/ado-aw/skills/.
    • macOS/Linux sandbox loopback/local-network restriction (v1.0.83): "sandboxed commands can no longer reach services running on your machine" by default, closing a local-service access path from within the sandbox.

    Notable features for ado-aw to adopt

    • MCP 2026-07-28 protocol support (v1.0.81): "Ship MCP 2026-07-28 support to CLI, SDK, IDE, and in-memory clients" — ado-aw's MCPG/MCP wiring should confirm compatibility with the newer protocol revision.
    • OTel trace context propagated into hooks (v1.0.81): hooks now receive traceparent/tracestate, letting hook-emitted spans correlate with the agent's trace — could improve ado-aw audit's OTel-based agent stats (src/agent_stats.rs) if ado-aw ever adds custom hooks.
    • Per-agent usage metrics in --usage-output-file (v1.0.81): richer usage breakdown that ado-aw audit's OTel/usage analyzer could surface per-subagent.
    • --with-token reads an auth token from stdin (v1.0.81): a non-interactive credential-injection path that may be relevant to how ado-aw wires the Copilot engine's auth in sandboxed pipeline steps.

    Deprecations

    • No formally-announced (not-yet-removed) deprecations were called out in this range beyond the renames/removals already listed under Breaking changes.

    Warning

    Firewall blocked 1 domain

    The following domain was blocked by the firewall during workflow execution:

    • index.crates.io

    To allow these domains, add them to the network.allowed list in your workflow frontmatter:

    network:
      allowed:
        - defaults
        - "index.crates.io"

    See Network Configuration for more information.

    Generated by Dependency Version Updater · auto · 217.5 AIC · ⌖ 14.6 AIC · ⊞ 14.3K · ◷

  6. github-actions commented on Sep 22, 2026

    @github-actions
    ContributorAuthor

    copilot-cli 1.0.70 → 1.0.87

    Releases analyzed

    (Releases v1.0.71–v1.0.81 and v1.0.84 either did not exist as public tags or were not returned by the release listing; the above are the releases actually present in the 1.0.70 < v <= 1.0.87 range.)

    Breaking changes

    • CLI argument parsing moved from Commander to a native Rust grammar (v1.0.85) — error and help text wording changed, copilot login --host now works, and --max-autopilot-continues no longer accepts scientific notation. ado-aw does not pass --max-autopilot-continues or parse copilot --help output programmatically, so no immediate action, but worth a quick check if any prompt/skill documentation quotes old CLI help text verbatim.
    • copilot plugins subcommand surface reorganized (v1.0.85): copilot plugins install --skill → copilot skill add; cross-kind --kind/--scope/--mcp/--skill flags removed from copilot plugins; copilot plugins list --json now emits a flat plugin array instead of { plugins, errors }; copilot plugins list is now an alias of copilot plugin list (plugins only, not MCP/skills/instructions/LSP). ado-aw doesn't invoke any copilot plugins/copilot plugin/copilot mcp/copilot skill subcommands in generated pipeline steps, so no action needed for the compiler itself — flagging in case any doc snippet in prompts/ or docs/ references the old copilot plugins install --skill form.
    • Sandboxed commands can no longer reach services on the host machine (macOS/Linux) (v1.0.83) — on macOS this also blocks a server the sandboxed command itself starts on 127.0.0.1. ado-aw runs the agent inside AWF's own network-isolation sandbox (not relying on Copilot CLI's own sandbox for network policy), so this shouldn't change AWF-mediated network behavior, but is worth confirming if any workflow relies on the agent binding a local port for internal tooling.
    • Linux sandboxing now requires slirp4netns, nsenter, iptables, ip6tables, iptables-restore, ip6tables-restore on PATH (v1.0.83) for proxy-mode network egress restriction. Relevant if the AWF/agent container image doesn't already provide these; worth a compatibility check against the images ado-aw pulls for the agent step.

    Security fixes

    • Secrets exported in the launching shell are no longer written to debug logs when a session is created or resumed (v1.0.87) — directly relevant to ado-aw's credential-isolation design; good confirmation that upstream is hardening the same threat model ado-aw's ado-proxy addresses.
    • Enterprise-denied MCP servers can no longer start before the managed allow/deny policy resolves (v1.0.83) — server startup now waits for managed-settings fetch instead of racing it, closing a policy-bypass race window.
    • allowManagedHooksOnly policy now also blocks extension-registered hook callbacks (preToolUse, postToolUse, postToolUseFailure) that previously bypassed managed-only lockdown (v1.0.85).

    Notable features for ado-aw to adopt

    • include-custom-instructions: true custom-agent frontmatter flag (v1.0.86) lets a custom agent opt into repository instruction files (AGENTS.md, copilot-instructions.md, CLAUDE.md). Could be surfaced as a front-matter passthrough for ado-aw agent files that want repo-level instructions automatically merged.
    • Client ID Metadata Document (CIMD) support for MCP OAuth sign-in (v1.0.83) — relevant if ado-aw ever wires user-defined MCP servers requiring OAuth.
    • Custom agents can list several model values, tried in order, with model-policy: required (v1.0.83) — a fallback-model list could be a useful engine.model extension for ado-aw's engine: front-matter.
    • Automatic HTTPS proxy mTLS client certificate support (v1.0.83) — relevant to any future corporate-proxy / --upstream-proxy AWF integration requiring client certs.
    • User/managed startup defaults for the Auto routing tier with strict/user-overridable org policy (v1.0.87) — worth reviewing for engine.model: auto behavior in ado-aw pipelines.
    • slowConnectionThresholdMs per-MCP-server configuration (v1.0.87) — could be exposed via ado-aw's MCP server config surface for tuning slow user-defined MCP servers.

    Deprecations

    • --no-auto-login and --no-sandbox remain supported but stay hidden from --help (v1.0.85) — unchanged behavior, just no longer documented in help output.
    • copilot plugins list/install --skill/cross-kind flags superseded by copilot plugin/copilot skill/copilot mcp equivalents (see Breaking changes above) — the old spellings still work as aliases per the release notes, but new documentation should prefer the new subcommand forms.

    Warning

    Firewall blocked 1 domain

    The following domain was blocked by the firewall during workflow execution:

    • index.crates.io

    To allow these domains, add them to the network.allowed list in your workflow frontmatter:

    network:
      allowed:
        - defaults
        - "index.crates.io"

    See Network Configuration for more information.

    Generated by Dependency Version Updater · auto · 211.2 AIC · ⌖ 4.95 AIC · ⊞ 14.3K · ◷

  7. github-actions commented on Sep 26, 2026

    @github-actions
    ContributorAuthor

    Consolidated into the canonical rolling issue #2232 ( [deps-release-notes] copilot-cli — upstream release action items ). Its action items have been preserved there.

    Generated by Deps Release-Notes Consolidator · auto · 74.6 AIC · ⌖ 4.15 AIC · ⊞ 9.6K

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

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions