Skip to content
DevAgentRadarDevAgentRadar
Sign in
Proposals/Configure GITHUB_TOKEN explicitly for Copilot CL
proposalteamP5High impactGCGitHub Copilot CLI

Configure GITHUB_TOKEN explicitly for Copilot CLI sandboxed shells

Copilot CLI v1.0.92-4 withholds ambient GITHUB_TOKEN from sandboxed shells unless you configure it, and adds copilot config list/read/set/remove. Upgrade, stop depending on inherited tokens, and manage settings through the new config commands.

Why this loop

v1.0.92-4 changes sandbox credential handling: sandboxed shells withhold ambient GITHUB_TOKEN unless explicitly configured, so agent tasks that previously inherited a token will lose GitHub API access until you grant it. The same release adds copilot config subcommands (list, read, set, remove) as the way to change settings, and copilot sandbox ca now honors --config-dir / -C so CA and CLI settings stay aligned. MCP startup is faster with many servers; legacy HTTP+SSE connections no longer hang when a POST is unacknowledged (bounded by the server timeout); MCP tools recover in the same turn when server instructions change. Sandboxed uv can write its cache when dev tool access is enabled, and pnpm no longer hits lock-file permission errors—keep package-manager work inside the sandbox. Compaction keeps the latest prompt when context is exceeded, so do not re-send the last user turn after a compact.

Proposed actions

  1. Upgrade GitHub Copilot CLI to v1.0.92-4 (https://github.com/github/copilot-cli/releases/tag/v1.0.92-4) and restart the CLI so sandbox token withholding, copilot config, and MCP timeout bounds are active.
  2. Run copilot config list; use copilot config read <setting>, copilot config set <setting> <value>, and copilot config remove <setting> instead of hand-editing Copilot CLI config files.
  3. If a sandboxed Copilot CLI session must call GitHub APIs, explicitly configure GITHUB_TOKEN for that sandbox; do not rely on ambient GITHUB_TOKEN inheritance (withheld unless configured in v1.0.92-4).
  4. When using a non-default config directory, run CA as copilot sandbox ca --config-dir <path> (also works with -C) so sandbox CA uses the same config dir as the CLI.
  5. Leave MCP servers connected when their instructions change—tools recover in the same turn; for legacy HTTP+SSE servers, rely on the server-configured acknowledgement timeout instead of killing hung POSTs.

Agent prompt

Paste into your agent or query via MCP (get_agent_prompt) — free, no extra AI cost

AGENTS.md / CLAUDE.md / GEMINI.md rule update

DevAgentRadar → Repo Harness Rule Patch

Goal: update our repository's permanent harness instructions based on this release.

Update the file that tool actually reads. Do not dump everything into one file.

  • Claude Code → CLAUDE.md plus .claude/ (skills, hooks, settings, agents, commands). It does not read AGENTS.md natively; start CLAUDE.md with @AGENTS.md.
  • Codex, Copilot, Cursor, Factory Droid, Grok Build, Roo Code, Goose, OpenCode, Amp, Zed, Aider → AGENTS.md.
  • Gemini CLI / Antigravity → GEMINI.md. AGENTS.md only if context.fileName is set.
  • Cursor glob-scoped rules → .cursor/rules/*.mdc (plain .md is ignored), not a second constitution.
  • Codex MCP → .codex/config.toml, not .mcp.json.
  • Copilot extra instructions → .github/copilot-instructions.md; custom agents → .github/agents/.
  • Windsurf → .windsurf/rules (do not assume AGENTS.md).
  • Cline → .clinerules.
  • Procedures → a skill (SKILL.md). Enforcement the model must not skip → a hook. Isolated roles → subagents.
  • Do not fork the same rule into three tool files.

Context

Assistant: GitHub Copilot CLI Proposal: Configure GITHUB_TOKEN explicitly for Copilot CLI sandboxed shells Summary: Copilot CLI v1.0.92-4 withholds ambient GITHUB_TOKEN from sandboxed shells unless you configure it, and adds copilot config list/read/set/remove. Upgrade, stop depending on inherited tokens, and manage settings through the new config commands. Primary source: https://github.com/github/copilot-cli/releases/tag/v1.0.92-4

Why it matters

v1.0.92-4 changes sandbox credential handling: sandboxed shells withhold ambient GITHUB_TOKEN unless explicitly configured, so agent tasks that previously inherited a token will lose GitHub API access until you grant it. The same release adds copilot config subcommands (list, read, set, remove) as the way to change settings, and copilot sandbox ca now honors --config-dir / -C so CA and CLI settings stay aligned. MCP startup is faster with many servers; legacy HTTP+SSE connections no longer hang when a POST is unacknowledged (bounded by the server timeout); MCP tools recover in the same turn when server instructions change. Sandboxed uv can write its cache when dev tool access is enabled, and pnpm no longer hits lock-file permission errors—keep package-manager work inside the sandbox. Compaction keeps the latest prompt when context is exceeded, so do not re-send the last user turn after a compact.

Suggested actions

  1. Upgrade GitHub Copilot CLI to v1.0.92-4 (https://github.com/github/copilot-cli/releases/tag/v1.0.92-4) and restart the CLI so sandbox token withholding, copilot config, and MCP timeout bounds are active.
  2. Run copilot config list; use copilot config read <setting>, copilot config set <setting> <value>, and copilot config remove <setting> instead of hand-editing Copilot CLI config files.
  3. If a sandboxed Copilot CLI session must call GitHub APIs, explicitly configure GITHUB_TOKEN for that sandbox; do not rely on ambient GITHUB_TOKEN inheritance (withheld unless configured in v1.0.92-4).
  4. When using a non-default config directory, run CA as copilot sandbox ca --config-dir <path> (also works with -C) so sandbox CA uses the same config dir as the CLI.
  5. Leave MCP servers connected when their instructions change—tools recover in the same turn; for legacy HTTP+SSE servers, rely on the server-configured acknowledgement timeout instead of killing hung POSTs.

Config surfaces this release may change

  • MCP servers (high confidence) — check your repo before applying
  • sandbox settings — check your repo before applying

After you finish

Do not report this as applied to DevAgentRadar. You cannot write the visitor's loop.

Tell the human: open https://devagentradar.com/proposals/github-copilot-cli-v1-0-92-4-configure-github-token-explicitly-for-copil and mark Applied, Skipped, or Failed. Proposal id: 55cb8b9d-7a72-4944-9661-e53b2f4680c0

Instructions:

  1. Read the existing context files and settings for the tools in this repo.
  2. Draft an explicit Git diff. Update ONLY agent harness rules or tool configuration.
  3. Call out deprecated flags, obsolete habits, or changed permission boundaries.
  4. Do not touch application logic.

Start with the proposed diff and say which layer and which tool's file it belongs in.

agentmcpmodelpricingsecurityRelease source ↗

Your loop

This browser · no sign-in · not shared as “you”

After you run the prompt

Only you can mark this. Agents cannot write your loop.

Your decision stays on this device. A public tally appears after a few votes.

Originating release signal

GCGitHub Copilot CLIv1.0.92-4Oct 4, 2026

v1.0.92-4 1.0.92-4

Added · Add copilot config subcommands to list, read, set, and remove settings. · Improved · Improve first-run startup by extracting the bundled CLI package in a child process · Improve startup responsiveness when connecting many MCP servers at once · Canvas actions can now return images to the model in invoke_canvas_action. · Fixed · Legacy HTTP+SSE MCP connections no longer hang indefinitely when a message POST is never acknowledged; the acknowledgement is bounded by the server's configured timeout · +20 more changes
Verified excerpt — the source's own words

Added

  • Add copilot config subcommands to list, read, set, and remove settings.

Improved

  • Improve first-run startup by extracting the bundled CLI package in a child process
  • Improve startup responsiveness when connecting many MCP servers at once
  • Canvas actions can now return images to the model in invoke_canvas_action.

Fixed

  • Legacy HTTP+SSE MCP connections no longer hang indefinitely when a message POST is never acknowledged; the acknowledgement is bounded by the server's configured timeout
  • Voice runtime install errors name why the download from nuget.org failed, not only the fallback feed's 401
  • Compaction keeps your latest prompt when requests exceed context limits
  • Large Anthropic requests rejected by provider size limits now retry after downscaling images or removing attachments
  • Custom agent model entries keep model-bound reasoning effort only when that model is selected
  • Shell tool calls now stream live stdout and stderr output reliably in the timeline
  • Usage reporting preserves provider-reported reasoning token totals when available
  • Sessions no longer slow to a crawl for minutes after the agent writes a very large file in one step

Excerpt ends here — this release continues at the source ↗.

Primary source ↗
Configure GITHUB_TOKEN explicitly for Copilot CLI sandboxed shells · DevAgentRadar