fix: preserve Claude stdio MCP env by imk1t · Pull Request #1224 · microsoft/apm

imk1t · 2026-05-09T09:03:04Z

Description

Preserve env values for self-defined stdio MCP servers when generating Claude Code .mcp.json.

Root cause: self-defined stdio MCP entries pass env into the shared adapter formatting path as a plain dictionary. The Copilot runtime-substitution path already handled that shape, but the legacy literal-resolution path used by Claude Code only handled registry-style environment variable lists, so the dictionary resolved to an empty object.

This keeps the fix scoped to that literal-resolution path and adds a Claude Code regression test proving that .mcp.json preserves the declared environment variable.

Fixes #1222

Type of change

Bug fix
New feature
Documentation
Maintenance / refactor

Testing

Tested locally
All existing tests pass
Added tests for new functionality (if applicable)

Validation run:

uv run --extra dev pytest tests/unit/test_claude_mcp.py tests/unit/test_build_mcp_entry.py tests/test_codex_empty_string_and_defaults.py -q
uv run --project <repo> --extra dev apm install --target claude --no-policy
uv run --extra dev pytest tests/unit tests/test_console.py -x
uv run --extra dev ruff check src/ tests/
uv run --extra dev ruff format --check src/ tests/

The CLI smoke test used a generic self-defined stdio MCP entry with DEMO_ENV=demo-value and verified that the generated .mcp.json contains:

"env": {
  "DEMO_ENV": "demo-value"
}

imk1t · 2026-05-09T15:07:34Z

@microsoft-github-policy-service agree

Copilot

Pull request overview

Fixes a Claude Code-specific regression where self-defined stdio MCP servers lost their declared env when generating .mcp.json, by teaching the legacy (literal-resolution) env resolver to handle the self-defined env dict shape.

Changes:

Extend _resolve_environment_variables to resolve env provided as a plain {NAME: value} dict when runtime env substitution is disabled (Claude Code path).
Add a unit regression test asserting that .mcp.json preserves env for a self-defined stdio server.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
`tests/unit/test_claude_mcp.py`	Adds regression coverage for preserving `env` in Claude Code `.mcp.json` for self-defined stdio servers.
`src/apm_cli/adapters/client/copilot.py`	Updates legacy env-variable resolution to support dict-shaped `env` inputs used by self-defined stdio MCP entries.

imk1t · 2026-05-10T09:39:12Z

+        if isinstance(env_vars, dict):
+            resolved = {}
+            for name, value in env_vars.items():
+                if not name:
+                    continue
+                if isinstance(value, str):
+                    resolved[name] = self._resolve_env_variable(
+                        name, value, env_overrides=env_overrides
+                    )
+                elif value is not None:
+                    resolved[name] = value
+            return resolved


github-actions · 2026-05-10T15:51:52Z

APM Review Panel: `needs_rework`

fix: preserve Claude stdio MCP env + normalize env values to strings -- fixes silent env loss for integer/float configs but ships a boolean coercion bug (False->"False") that needs a one-liner before merge

cc @danielmeppiel -- a fresh advisory pass is ready for your review.

The core fix is correct and should ship. This PR closes a real, high-frequency user pain: any apm.yml that sets numeric env values (PORT: 3000, TIMEOUT: 30) silently produced a broken Claude stdio MCP config. The new legacy-mode dict branch correctly routes non-string values into a str() coercion path and adds a regression test for the integer case. The panel converges on shipping this direction -- the fix is the right call and the code placement is sound.

However, the PR introduces a secondary correctness bug that the test suite does not catch: str(False) == "False" (capital F). DevX UX flagged this blocking, supply-chain confirmed it, and the test-coverage expert surfaced it as a missing-test gap. Shell tooling and 12-factor apps universally expect lowercase "false"/"true". This is not hypothetical -- any user with DEBUG: false or ENABLED: true in apm.yml gets a semantically wrong string baked into ~/.claude.json. The fix is a one-liner (str(value).lower() when isinstance(value, bool)) and must land in the same commit, not a follow-up. The panel also independently identifies the translate-mode dict branch (line 748-749) as inconsistent: it does not apply the same str() coercion, meaning the same input shape produces different output types depending on which code path runs. This asymmetry is a latent bug. The CHANGELOG omission is flagged by three personas (doc-writer, devx-ux, oss-growth) -- per project convention a user-visible bug fix requires a changelog entry before merge.

The print() on line 819 flagged blocking by cli-logging is pre-existing and not introduced by this PR. It should not gate merge; track it as a standalone cleanup issue. The supply-chain concern about str() on YAML list/dict values producing Python repr is real but low-urgency -- an explicit type whitelist is a reasonable follow-up.

Dissent. cli-logging-expert rated the print() on line 819 as blocking. This finding is correct on the merits -- raw print() violates the APM output architecture -- but the line is pre-existing and not introduced by this PR. Blocking a focused bug-fix PR on a pre-existing unrelated violation is disproportionate; the CEO sides with not gating merge on it.

Aligned with: Pragmatic as npm: numeric YAML env values (PORT, TIMEOUT, DEBUG) are the natural way humans write config -- the fix meets users where they are rather than demanding string quoting as a workaround. Portable by manifest: env normalization must be deterministic across all code paths; the translate-mode branch inconsistency (line 748-749) breaks this promise for the same apm.yml input. Secure by default: YAML list/dict values producing Python repr in a config-bake path is a reliability risk worth a follow-up type whitelist. DevX: boolean env values are a trap (str(False) == "False"); lowercase coercion is the correct UX contract and must ship with this fix.

Growth signal. Community contributor Koichi Takahashi fixed a silent failure that would hit any user configuring stdio MCP servers with numeric YAML env values -- an extremely common pattern (PORT, DEBUG, TIMEOUT). The next release post should call this out explicitly as a community-contributed fix. Consider adding a working stdio-with-numeric-env example to the docs quickstart to prevent future confusion and convert this fix into a discovery surface for new users.

Panel summary

Persona	B	R	N	Takeaway
Python Architect	0	1	1	New dict-env branch is correctly placed and scoped; one type-consistency gap between translate-mode and legacy-mode dict handling; docstring needs updating.
CLI Logging Expert	1	2	1	Silent coercion of non-string env values needs a verbose-mode diagnostic; bare print() on line 819 violates STATUS_SYMBOLS convention; translate-mode dict path inconsistently leaves non-strings as-is.
DevX UX Expert	1	2	1	str() coercion silently capitalises YAML booleans (false->False); None silently dropped; CHANGELOG entry missing; test coverage thin on edge cases.
Supply Chain Security Expert	0	1	2	No blocking supply-chain issues; new dict-env path is consistent with legacy bake-to-disk behavior; one recommended note on str() of complex YAML values.
OSS Growth Hacker	0	1	1	Good external-contributor fix for a high-frequency silent failure (PORT as int breaks Claude); CHANGELOG entry is absent -- blocks users from knowing the fix shipped.
Doc Writer	0	1	1	CHANGELOG [Unreleased] is missing the Claude stdio env fix entry; existing docs describe intended behavior correctly so no doc drift found.
Test Coverage Expert	0	1	1	Int-to-string coercion is tested (PORT:3000->"3000"); bool coercion (False->"False") and None-value silent-drop have no regression trap; unit tier is correct for this pure function.

B = blocking-severity findings, R = recommended, N = nits.
Counts are signal strength, not gates. The maintainer ships.

Top 5 follow-ups

[DevX UX Expert] (blocking-severity) Boolean env values coerce to 'False'/'True' (capital) instead of 'false'/'true' -- fix in same PR before merge -- str(False)=="False" is a correctness bug introduced by this PR. Missing test confirmed by test-coverage-expert (evidence: outcome=missing). One-liner fix: str(value).lower() when isinstance(value, bool). Must land before merge.
[Python Architect] Translate-mode dict branch (line 748-749) does not apply str() coercion -- same input, different output type depending on code path -- independently flagged by both python-architect and cli-logging-expert. Breaks the dict[str,str] return-type contract. One-line fix mirrors what this PR already does in the legacy branch.
[OSS Growth Hacker] Add CHANGELOG [Unreleased] ### Fixed entry for the stdio env normalization fix before merge -- three personas converged on this (doc-writer, devx-ux, oss-growth). Users upgrading cannot discover the fix without it. Per project convention every user-visible bug fix requires a changelog bullet.
[Supply Chain Security Expert] Add explicit type whitelist (int/float/bool) for str() coercion; warn-and-skip YAML list/dict values -- str(["a","b"]) produces Python repr baked into ~/.claude.json. A whitelist prevents confusing runtime failures and makes the contract explicit.
[Test Coverage Expert] Add None-value-in-dict regression test to pin the intentional silent-drop behavior -- the None-drop is currently undocumented in the test suite. A test asserting "ABSENT" not in env makes the contract explicit and prevents a future change from silently altering behavior.

Architecture

classDiagram
    direction LR

    class MCPClientAdapter {
        <<AbstractBase>>
        +configure_mcp_server(ref) bool
    }

    class CopilotClientAdapter {
        <<ConcreteAdapter>>
        +_supports_runtime_env_substitution bool
        +_last_env_placeholder_keys set
        +_last_legacy_angle_vars set
        +_resolve_environment_variables(env_vars, env_overrides) dict
        +_resolve_env_variable(name, value, env_overrides) str
        +configure_mcp_server(ref) bool
    }

    class EnvVars_DictInput {
        <<ValueObject>>
        +name str
        +value str or int or float or None
    }

    class EnvVars_ListInput {
        <<ValueObject>>
        +name str
        +description str
        +required bool
    }

    class ResolvedEnv {
        <<ValueObject>>
        +name str
        +value str
    }

    MCPClientAdapter <|-- CopilotClientAdapter : extends
    CopilotClientAdapter ..> EnvVars_DictInput : accepts
    CopilotClientAdapter ..> EnvVars_ListInput : accepts
    CopilotClientAdapter ..> ResolvedEnv : returns

    class CopilotClientAdapter:::touched
    classDef touched fill:#fff3b0,stroke:#d47600

flowchart TD
    A["_resolve_environment_variables(env_vars, env_overrides)"] --> B{"isinstance(env_vars, dict)?"}

    B -- "yes" --> C{"_supports_runtime_env_substitution?"}
    B -- "no" --> F{"_supports_runtime_env_substitution?"}

    C -- "yes (translate mode)" --> D["Branch 1: translate placeholders\nline 742-766"]
    D --> D1{"isinstance(raw_value, str)?"}
    D1 -- "no" --> D2["[BUG] translated[name] = raw_value as-is\nnon-string, may break JSON contract\nline 748-750"]
    D1 -- "yes + placeholder" --> D3["_translate_env_placeholder(raw_value)"]
    D1 -- "yes, literal default" --> D4["raw_value kept literal"]
    D1 -- "yes, other literal" --> D5["dollar-brace NAME (secret guard)"]

    C -- "no (legacy dict, NEW branch)" --> E["Branch 3 [NEW]: coerce dict env\nline 791-802"]
    E --> E1{"isinstance(value, str)?"}
    E1 -- "yes" --> E2["_resolve_env_variable(name, value, env_overrides)"]
    E1 -- "no and not None" --> E3["[FIX] str(value)\n[FOLLOW-UP: bool needs .lower()]"]
    E1 -- "None" --> E4["skip key"]

    F -- "yes" --> G["Branch 2: emit dollar-brace placeholders\nlist input, line 768-789"]
    F -- "no" --> H["Branch 4: legacy list + Rich prompts\nline 804-868"]

Recommendation

The core fix is correct and should ship. Do not revert or redesign. But the PR needs three targeted changes before merge: (1) fix the boolean coercion bug (str(value).lower() when isinstance(value, bool)) and add a parametrized test covering bool cases -- this is a correctness regression introduced by the PR itself, confirmed missing by the test-coverage expert; (2) apply the same str() coercion to the translate-mode dict branch at line 748-749 to restore type-contract consistency across both paths; (3) add a CHANGELOG [Unreleased] ### Fixed bullet with a user-symptom-first description. All three changes are small (combined under 15 lines) and should be done in the same PR before merge. The print() on line 819 is pre-existing and must not gate this fix; file it as a separate issue.

Full per-persona findings

Python Architect

[recommended] Translate-mode dict branch passes non-string values through as raw Python objects; new legacy branch coerces to str -- inconsistent return-type contract at src/apm_cli/adapters/client/copilot.py:748
At line 748-750, the translate-mode dict branch does translated[name] = raw_value; continue for any non-string value. The new legacy dict branch (line 800-801) does str(value) for the same input shape. Both branches should return dict[str, str] but the translate branch can return dict[str, int|float|...].
Suggested: Change translated[name] = raw_value to translated[name] = str(raw_value) in the translate-mode dict branch (line 749), mirroring the fix applied here.
[nit] Docstring Args section still says env_vars (list) after the fix, but the function now accepts both list and dict at src/apm_cli/adapters/client/copilot.py:720
The Args block documents env_vars as (list) only. This PR adds a third dict-specific branch, making the dual-type contract more prominent.
Suggested: Change env_vars (list) to env_vars (list | dict) and add a second bullet explaining the dict shape.

CLI Logging Expert

[blocking] print() on line 819 bypasses _rich_* / STATUS_SYMBOLS convention and leaks test-infra noise to stdout at src/apm_cli/adapters/client/copilot.py:819
Raw print() calls violate the APM output architecture. This specific line also uses an f-string with no interpolation (noqa:F541 comment confirms it). Note: this is pre-existing, not introduced by this PR.
Suggested: Replace with _rich_info('[i] APM_E2E_TESTS detected, skipping environment variable prompts') or gate behind verbose flag.
[recommended] Silent str() coercion of non-string env values emits no diagnostic -- user cannot tell PORT:3000 became string '3000' at src/apm_cli/adapters/client/copilot.py:801
The new dict+non-translate path silently coerces any non-string value. Under --verbose this should emit at minimum a dim info line so power users can audit what was baked into config.
Suggested: After resolved[name] = str(value), add a verbose-guarded log: [i] ENV: NAME coerced type->str.
[recommended] Translate-mode dict path (line 749) preserves non-string values as raw Python objects -- inconsistent with the new non-translate fix at src/apm_cli/adapters/client/copilot.py:749
PORT:3000 in translate mode is stored as int, while the new non-translate path coerces to str. Two branches have opposite behaviour for the same input shape.
Suggested: Replace translated[name] = raw_value with translated[name] = str(raw_value).
[nit] Docstring for _resolve_environment_variables still says env_vars is list at src/apm_cli/adapters/client/copilot.py:720
Suggested: Update to env_vars (list | dict).

DevX UX Expert

[blocking] YAML boolean false coerces to 'False' (capital F), not 'false' at src/apm_cli/adapters/client/copilot.py:801
Python str(False) == 'False'. A user writing DEBUG: false in apm.yml gets env var value 'False', but every shell tool and 12-factor app expects lowercase 'false'/'true'. Silent semantic breakage worse than the original int bug.
Suggested: Use str(value).lower() if isinstance(value, bool) else str(value). Add a test.
[recommended] None values are silently dropped with no user feedback at src/apm_cli/adapters/client/copilot.py:800
A key present in apm.yml but set to null is skipped without any warning. The user sees the server installed successfully but the env key is absent from ~/.claude.json.
Suggested: Emit a warning: 'env var PORT has null value in apm.yml and will be omitted from the MCP config'.
[recommended] CHANGELOG [Unreleased] ### Fixed section has no entry for this fix at CHANGELOG.md
Users upgrading who had broken apm.yml with integer env values cannot discover this fix. Every user-visible bug fix must have a changelog entry.
Suggested: Add bullet: stdio MCP servers with non-string env values (e.g. PORT: 3000) now write correctly as strings to ~/.claude.json / .mcp.json. (#1222)
[nit] Test only covers int coercion; bool and float cases untested at tests/unit/test_claude_mcp.py:129
Suggested: Extend test with DEBUG: False, RATE: 0.5, TIMEOUT: 0 assertions.

Supply Chain Security Expert

[recommended] str(value) on a YAML list or nested dict produces Python repr, not a usable env value at src/apm_cli/adapters/client/copilot.py:801
If a user writes env: {FOO: [a, b]}, YAML parses it as a Python list. str(['a','b']) yields "['a', 'b']" (Python repr) baked into ~/.claude.json. Not an injection risk but silently incorrect.
Suggested: Use isinstance(value, (int, float, bool)) whitelist; warn-and-skip list/dict types.
[nit] bool coercion: str(True)=='True', str(False)=='False' (capital) diverges from YAML/JSON convention at src/apm_cli/adapters/client/copilot.py:801
Suggested: Use str(value).lower() when isinstance(value, bool).
[nit] Non-string literal values bypass _resolve_env_variable -- add inline comment to explain this is intentional at src/apm_cli/adapters/client/copilot.py:800
Suggested: Add comment: # Non-string literals cannot contain ${VAR} placeholders; coerce directly.

OSS Growth Hacker

[recommended] No CHANGELOG entry exists for the fix
Numeric env values (PORT, DEBUG=1, TIMEOUT=30) are the most natural YAML you write; this bug silently broke every Claude stdio MCP config with non-string env values. A high-visibility user-facing fix MUST appear in [Unreleased] ### Fixed.
Suggested: Add: stdio MCP servers declared with plain-dict env: blocks (e.g. PORT: 3000) now work correctly with Claude: non-string values are coerced to strings. (#1222) -- by @koichirok``
[nit] CHANGELOG entry should lead with user symptom not code jargon
Suggested: Lead with: If your apm.yml sets numeric env values (e.g. PORT: 3000), Claude now receives them as strings instead of silently dropping them.

Doc Writer

[recommended] CHANGELOG [Unreleased] ### Fixed section is missing an entry for this PR's bug fix at CHANGELOG.md
The fix is a user-visible bug fix. Per project convention every merged fix gets a [Fixed] CHANGELOG bullet. The PR's rebase activity removed unrelated entries but never added one for its own change.
Suggested: Add to [Unreleased] ### Fixed: apm install --target claude now preserves env entries on stdio MCP servers and normalizes all env values to strings, fixing silent env loss for Claude stdio MCP server configs. (#1222)
[nit] dependencies.md says Claude Desktop but adapter targets Claude Code (.claude/) -- label may need a follow-up clarification at docs/src/content/docs/guides/dependencies.md
Suggested: Verify Claude Desktop vs Claude Code labeling; update to Claude Code if appropriate.

Auth Expert -- inactive

The PR only touches copilot.py env-var normalization (non-string MCP values), test_claude_mcp.py, and CHANGELOG.md -- none of which involve token management, credential resolution, AuthResolver, or remote-host auth semantics.

Test Coverage Expert

[recommended] No test for bool env value coercion: False -> "False" is non-obvious and untested at tests/unit/test_claude_mcp.py
For a bool False, str() produces "False" (capital F). Many downstream tools expect lowercase 'false'. No test covers the bool case. Grepped tests/ -- no hit for bool dict env in the new code path.
Suggested: Add parametrized case: {"VERBOSE": False, "RETRIES": True} -> {"VERBOSE": "False", "RETRIES": "True"} (pins current contract; catches any future normalisation change).
Proof (missing): tests/unit/test_claude_mcp.py::test_configure_self_defined_stdio_coerces_bool_env_to_string -- proves: When a server declares a boolean env value, apm install writes the Python str() form to .mcp.json [devx]
[nit] None-valued env key is silently dropped; no test asserts the drop is intentional at tests/unit/test_claude_mcp.py
The legacy dict branch skips any key whose value is None. Undocumented in the test suite. Grepped all tests -- no test exercises the None-value-in-dict case.
Suggested: Add test: {"PRESENT": "val", "ABSENT": None} -> expected {"PRESENT": "val"} (ABSENT key absent).
Proof (missing): tests/unit/test_claude_mcp.py::test_configure_self_defined_stdio_drops_none_env_values -- proves: When env dict contains a None-valued key, apm install omits that key from the written .mcp.json

_{This panel is advisory. It does not block merge. Re-apply the

panel-review label after addressing feedback to re-run.}

Note

🔒 Integrity filter blocked 2 items

The following items were blocked because they don't meet the GitHub integrity level.

#1224 pull_request_read: has lower integrity than agent requires. The agent cannot read data with integrity below "approved".
fix: preserve Claude stdio MCP env #1224 pull_request_read: has lower integrity than agent requires. The agent cannot read data with integrity below "approved".

To allow these resources, lower min-integrity in your GitHub frontmatter:

tools:
  github:
    min-integrity: approved  # merged | approved | unapproved | none

Generated by PR Review Panel for issue #1224 · ● 3.3M · ◷

imk1t · 2026-05-11T11:55:44Z

@danielmeppiel @hansonkim
If you would like, it is possible to address #1266 as well in this PR.

Address microsoft#1224 PR Review Panel feedback from issuecomment-4415698348.

…MCP (closes microsoft#1266) The Codex and Gemini adapters wrote the env block of self-defined stdio MCP servers to disk verbatim, bypassing the env-var resolution pipeline. All three placeholder forms (${VAR}, ${env:VAR}, legacy <VAR>) landed as literal strings in ~/.codex/config.toml and .gemini/settings.json, so the MCP child processes received e.g. ATLASSIAN_API_TOKEN=${ATLASSIAN_API_TOKEN} and authentication failed. Root cause was twofold. The Codex and Gemini _raw_stdio branches never called _resolve_environment_variables. Even adapters that did call it (Cursor) hit a latent gap: the legacy-mode branch only handled the registry list-of-dict shape, so the dict-shape input from self-defined stdio deps was iterated as keys, every key failed isinstance(..., dict), and the env block came out as {}. Codex additionally extends MCPClientAdapter directly, not CopilotClientAdapter, so the resolver method wasn't even available on the class. Fix moves _resolve_environment_variables, _resolve_env_variable, _resolve_variable_placeholders, the env-placeholder regex helpers, and the per-server placeholder state to MCPClientAdapter so every adapter shares one resolver contract. Adds a legacy-mode dict-shape branch (matches the fix in microsoft#1224 for Claude) and rewrites Codex/Gemini's _raw_stdio branches to route env and args through the resolver. Cursor's silent-drop latent bug dies as a byproduct of consolidating the resolver onto the base; pinned by a new regression test.

imk1t added 2 commits May 9, 2026 18:00

fix: preserve Claude stdio MCP env

29aab4e

Merge branch 'main' into fix/claude-stdio-env-1222

1b1a721

imk1t marked this pull request as ready for review May 9, 2026 15:07

imk1t requested a review from danielmeppiel as a code owner May 9, 2026 15:07

Copilot AI review requested due to automatic review settings May 9, 2026 15:07

imk1t requested a review from sergio-sisternes-epam as a code owner May 9, 2026 15:07

Copilot started reviewing on behalf of imk1t May 9, 2026 15:07 View session

Copilot AI reviewed May 9, 2026

View reviewed changes

imk1t added 2 commits May 10, 2026 01:13

fix: normalize Claude stdio env values to strings

33c8cf0

Merge branch 'main' into fix/claude-stdio-env-1222

162d358

This was referenced May 10, 2026

[Test Improver] tests(compile): add unit tests for compile() command early-exit paths #1233

Closed

[Test Improver] Monthly Activity 2026-05 #1101

Closed

danielmeppiel added the panel-review Trigger the apm-review-panel gh-aw workflow label May 10, 2026

github-actions Bot removed the panel-review Trigger the apm-review-panel gh-aw workflow label May 10, 2026

hansonkim mentioned this pull request May 11, 2026

[BUG] Codex and Gemini adapters pass self-defined stdio env verbatim, skipping placeholder resolution #1266

Open

edenfunf mentioned this pull request May 11, 2026

fix: resolve env-var placeholders in Codex/Gemini self-defined stdio MCP (closes #1266) #1277

Open

4 tasks

This was referenced May 12, 2026

[Test Improver] tests(install): add unit tests for integration template module #1287

Closed

[Test Improver] tests: skip runtime factory tests when LLM not installed #1139

Closed

Merge branch 'main' into fix/claude-stdio-env-1222

2e0ade2

danielmeppiel mentioned this pull request May 13, 2026

[Test Improver] tests(cache): add unit tests for cache.integrity SHA verification #1302

Closed

sergio-sisternes-epam added panel-review Trigger the apm-review-panel gh-aw workflow status/triaged Initial agentic triage complete; pending maintainer ratification (silence = approval). and removed panel-review Trigger the apm-review-panel gh-aw workflow labels May 13, 2026

imk1t and others added 2 commits May 14, 2026 08:48

fix: normalize dict-shaped MCP env literals

7fde972

Address microsoft#1224 PR Review Panel feedback from issuecomment-4415698348.

Merge branch 'main' into fix/claude-stdio-env-1222

e114838

danielmeppiel merged commit 894d73a into microsoft:main May 14, 2026
8 checks passed

danielmeppiel mentioned this pull request May 15, 2026

[docs] Update documentation for features from 2026-05-14 #1338

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fix: preserve Claude stdio MCP env#1224

fix: preserve Claude stdio MCP env#1224
danielmeppiel merged 7 commits into
microsoft:mainfrom
imk1t:fix/claude-stdio-env-1222

imk1t commented May 9, 2026 •

edited

Loading

Uh oh!

imk1t commented May 9, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

imk1t May 10, 2026

Uh oh!

github-actions Bot commented May 10, 2026

Python Architect

CLI Logging Expert

DevX UX Expert

Supply Chain Security Expert

OSS Growth Hacker

Doc Writer

Auth Expert -- inactive

Test Coverage Expert

Uh oh!

imk1t commented May 11, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Conversation

imk1t commented May 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Type of change

Testing

Uh oh!

imk1t commented May 9, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

imk1t May 10, 2026

Choose a reason for hiding this comment

Uh oh!

github-actions Bot commented May 10, 2026

APM Review Panel: needs_rework

Panel summary

Top 5 follow-ups

Architecture

Recommendation

Python Architect

CLI Logging Expert

DevX UX Expert

Supply Chain Security Expert

OSS Growth Hacker

Doc Writer

Auth Expert -- inactive

Test Coverage Expert

Uh oh!

imk1t commented May 11, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

imk1t commented May 9, 2026 •

edited

Loading

APM Review Panel: `needs_rework`