feat(discovery): comprehensive brownfield discovery, migration, and cross-tool skill install (#1122)

[chkp-roniz]

Summary

Implements the brownfield discovery feature (apm init --discover) requested in #1122, enabling cross-tool migration of agent context files.

What it does

• Discovery: Scans all known tool directories (.claude/, .codex/, .github/, .agents/, etc.) for instructions, agents, skills, commands, hooks, and styles at project, user, and system scope
• Migration: --write copies convertible files into canonical .apm/ locations with proper APM extensions (e.g. .claude/agents/helper.md -> .apm/agents/helper.agent.md)
• Skill wrapping: Plain .md skill files (Claude format) are wrapped into <name>/SKILL.md directory structures so the install pipeline can deploy them to all targets
• Includes tracking: All discovered dot dirs (.claude/, .codex/, .github/) are merged into apm.yml includes
• Cross-tool install: After discovery+migration, apm install --target <tool> deploys everything to the target tool

End-to-end scenario

# Developer has a Claude project, colleague uses Codex
apm init --discover --write   # scans .claude/, migrates to .apm/
apm install --target codex    # deploys to .codex/
# Colleague resumes work with Codex

Changes

New files

• src/apm_cli/context_discovery.py -- discovery engine, migration plan, manifest builder
• tests/unit/test_context_discovery.py -- 40 unit tests
• tests/integration/test_discover_e2e.py -- integration tests

Modified files

• src/apm_cli/commands/init.py -- --discover, --write, --format flags
• src/apm_cli/commands/compile/cli.py -- pre-flight check scans includes dirs
• src/apm_cli/primitives/parser.py -- normalize applyTo list values
• CHANGELOG.md -- 8 new entries (2 Added, 6 Fixed)
• docs/src/content/docs/reference/cli-commands.md -- migration and includes docs
• packages/apm-guide/.apm/skills/apm-usage/commands.md -- updated --discover row

Tool coverage

Tool	Rules	Migration
Claude	root-instructions, agent, command, skill (plain .md + SKILL.md), settings	Yes
Codex	root-instructions, agent, command, skill, instruction, settings	Yes
Copilot	instruction, skill, agent	Yes
Cursor	instruction, agent, command, skill	Yes
OpenCode	instruction, agent, command, skill	Yes
Gemini	instruction, agent, command, skill	Yes
Windsurf	instruction, agent, command, skill	Yes
APM native	all primitive types	Yes (misplaced files)

Closes #1122

[Copilot]

Pull request overview

Adds brownfield agent context discovery to apm init via --discover, enabling preview-first inventory of existing tool context (project/user/system scopes), optional migration into canonical .apm/ locations, and cross-tool install compatibility.

Changes:

• Introduces a discovery/migration engine (context_discovery.py) that classifies findings and computes an idempotent migration plan into .apm/.
• Extends apm init with --discover, --write, and --format output options, plus unit/integration coverage for preview/write flows.
• Updates compilation preflight checks, primitive parsing (applyTo normalization), docs, and changelog entries to reflect the new workflow.

Reviewed changes

Copilot reviewed 10 out of 10 changed files in this pull request and generated 9 comments.

Show a summary per file

File	Description
src/apm_cli/context_discovery.py	New discovery engine: ruleset scanning, findings/manifest output formatting, and migration planning/execution.
src/apm_cli/commands/init.py	Adds --discover/--write/--format flags and discovery-mode execution path.
src/apm_cli/commands/compile/cli.py	Preflight check enhancement to consider includes directories when determining local content presence.
src/apm_cli/primitives/parser.py	Normalizes applyTo frontmatter when YAML provides a list instead of a string.
tests/unit/test_context_discovery.py	New unit test suite validating discovery classifications, redaction, and migration behavior.
tests/unit/test_init_command.py	Adds apm init --discover tests for preview/write and structured output formats.
tests/integration/test_discover_e2e.py	New sanity E2E coverage for discovery preview, write mode, and JSON output.
docs/src/content/docs/reference/cli-commands.md	Documents apm init --discover flags, behavior, and examples.
packages/apm-guide/.apm/skills/apm-usage/commands.md	Updates command reference skill with the new init --discover row.
CHANGELOG.md	Adds discovery-related changelog entries (but currently has structure/placement issues).

[github-actions]

APM Review Panel: ship_with_followups

apm init --discover [--write] scans 7 AI tool configs and proposes or executes a full brownfield migration to apm.yml + .apm/ -- the single highest-leverage adoption unlock APM has shipped.

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

This PR delivers the first zero-to-one brownfield migration surface: a developer already using Claude, Cursor, Codex, Copilot, Gemini, OpenCode, or Windsurf can run two commands and land on APM without manual config rewriting. The architecture panel found no blocking correctness bugs -- the triple-collision issue in compute_migration_plan() is a real edge-case defect but affects only the preview display, not file safety, and is straightforwardly fixed. The test-coverage panel surfaced two missing-test gaps (_normalize_apply_to list-unwrap and compile pre-flight includes-dir); neither is a regression against the new feature's main path, but the compile pre-flight gap is a latent risk for users coming from the discover/write flow and should be tracked. The security panel's two recommended findings -- shutil.copytree symlink following and verbatim credential copying to .apm/ -- are the highest-stakes items in the set. Both deserve a follow-up issue with short-cycle resolution, and the credential scan is worth a pre-1.0 gate.

The oss-growth-hacker raised 'blocking' severity on the README omission of the migration story. Under the advisory panel contract this is not a merge gate, but it is the highest-signal positioning finding in the review: every AI developer who visits the repo already has tool-specific configs, and the two-command migration sequence is the most compelling first-sentence reason to try APM. This should land in the same release window. The devx-ux-expert's flag rename (--write -> --apply) is a clean mental-model improvement aligned with terraform plan/apply conventions; it is a breaking change to the CLI surface and should ship before this feature is widely adopted. The --format silent-ignore guard is a low-cost correctness fix that protects scripted consumers and should accompany the rename. CLI logging findings (wrong header on --write path, migration count divergence, wrong symbol for cancellation) affect trust in automated workflows and deserve a single cleanup commit.

Dissent. No direct disagreement between panelists. The oss-growth-hacker classified the README omission as 'blocking' while all other panelists' blocking-tier finding lists were empty. This is a framing difference (growth/positioning urgency vs. correctness/security urgency), not a substantive conflict. CEO sides with treating it as the highest-priority recommended action -- not a merge gate, but not deferrable past the release window.

Aligned with: Portable by manifest (existing tool-specific configs consolidated into a single apm.yml). Secure by default (symlink skipping and .git exclusion are in place; copytree symlink dereferencing and credential copying are open gaps). Multi-harness/multi-host (7-tool coverage is the broadest APM has shipped). Pragmatic as npm (preview-then-apply pattern matches modern package manager ergonomics).

Growth signal. PR #1227 is the first zero-to-one brownfield migration surface in APM's history. Growth priorities: (1) README hero update with a 'Already using Claude / Cursor / Copilot?' migration hook immediately after the install snippet, naming all 7 covered tools; (2) release post headline 'Migrate Claude, Cursor, or Copilot to APM in two commands'; (3) 7-tool coverage is a credible and defensible moat claim -- lead with the number; (4) use 'migrate' framing (not 'discover') in all user-facing copy. This PR directly addresses the single most common adoption objection from existing AI tool users.

Panel summary

Persona	B	R	N	Takeaway
Python Architect	0	2	4	context_discovery.py is well-structured with good frozen-dataclass usage; two concrete issues: a triple-collision bug in compute_migration_plan() and a pointless deep-copy of TOOL_CONTEXT_RULES into USER_RULES.
CLI Logging Expert	0	5	2	Text report bypasses CommandLogger entirely and has several signal/label issues; no blocking terminal breaks, but five recommended fixes before merge.
DevX UX Expert	0	2	2	Preview-first pattern is strong and well-documented; two naming issues (--write, --format coupling) will confuse users who read apm init --help out of context.
Supply Chain Security Expert	0	2	2	No blocking exploitable vulnerabilities; one recommended fix for symlink-following in shutil.copytree on skill directories, plus minor redaction and binary-detection notes.
OSS Growth Hacker	1	2	2	Zero-to-one brownfield migration story is the strongest adoption unlock APM has shipped, but it is completely absent from the README and the verb 'discover' obscures the user's actual intent ('migrate').
Doc Writer	0	3	3	--write flag description understates its file-migration scope; importability categories are referenced but never defined; several CHANGELOG Fixed entries describe new behavior rather than fixes.
Test Coverage Expert	0	2	0	Two coverage gaps: (1) _normalize_apply_to list-unwrap logic has no unit test; (2) compile pre-flight includes-dir path has no test at any tier.

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

Top 5 follow-ups

1. [OSS Growth Hacker] Add brownfield migration hook to README hero section -- two-command sequence, 7 tools named, placed immediately after install snippet -- The migration story is the Why do we need a GitHub token? #1 reason an existing Claude/Cursor/Codex user installs APM. Its absence from the README means the strongest adoption signal in this release is invisible to every cold visitor. Panelist rated blocking; advisory panel treats it as highest-urgency recommended action for same release window.
2. [Supply Chain Security Expert] shutil.copytree called without symlinks=True -- dereferences symlinks inside copied skill directories, can pull unexpected file contents into .apm/ -- Concrete footgun on adversarial skill repos. Pass symlinks=True or add pre-copy symlink scan. Straightforward fix with no user-visible behavioral change for normal repos.
3. [Supply Chain Security Expert] Credential-bearing source files are copied verbatim to .apm/ -- token redaction is display-only -- If a scanned file contains a GitHub PAT or embedded credentials, shutil.copy2 writes raw bytes into .apm/ and those secrets become tracked. Needs a pre-copy token pattern scan with warn-and-refuse or warn-and-redact path.
4. [DevX UX Expert] Rename --write to --apply (terraform plan/apply mental model); add guard so --format without --discover errors instead of silently no-ops -- --write is ambiguous standalone and will mislead users scanning apm init --help. --apply is the established convention. Both are breaking changes to the CLI surface -- cheapest to ship now before adoption widens.
5. [Test Coverage Expert] Add unit test for _normalize_apply_to list-unwrap and integration test for compile pre-flight includes-dir content probe -- The compile pre-flight gap is a latent regression trap: a future refactor could cause apm compile to abort for every project produced by the discover/write flow, with no automated guardrail to catch it. (evidence.outcome: missing on both.)

Architecture

classDiagram
    direction TB

    class ContextDiscoveryRule {
        <<ValueObject>>
        +tool: str
        +kind: str
        +patterns: tuple
        +importability: str
        +reason: str
    }

    class ContextDiscoveryFinding {
        <<ValueObject>>
        +path: Path
        +display_path: str
        +scope: str
        +tool: str
        +kind: str
        +importability: str
        +reason: str
        +to_dict() dict
    }

    class MigrationAction {
        <<ValueObject>>
        +source: Path
        +dest: Path
        +tool: str
        +kind: str
        +is_dir: bool
        +wrap_as_skill: bool
    }

    class ContextDiscoveryResult {
        <<ValueObject>>
        +project_root: Path
        +detected_target: str
        +existing_apm_yml: bool
        +findings: tuple
        +proposed_manifest: dict
        +migration_plan: tuple
        +summary() dict
        +to_dict() dict
    }

    class _MIGRATION_MAP {
        <<TableDispatch>>
        +key: tuple[tool, kind]
        +value: tuple[subdir, ext]
    }

    ContextDiscoveryResult *-- ContextDiscoveryFinding : findings tuple
    ContextDiscoveryResult *-- MigrationAction : migration_plan tuple
    ContextDiscoveryFinding ..> ContextDiscoveryRule : matched by _scan_scope
    _MIGRATION_MAP ..> MigrationAction : drives compute_migration_plan

    class ContextDiscoveryResult:::touched
    class ContextDiscoveryFinding:::touched
    class ContextDiscoveryRule:::touched
    class MigrationAction:::touched
    class _MIGRATION_MAP:::touched

    classDef touched fill:#fff3b0,stroke:#d47600

Loading

flowchart TD
    A(["apm init --discover --write"]) --> B["init.py: _run_discovery_init()"]
    B --> C["discover_agent_context()\nresolves project_root"]
    C --> D["_scan_scope(project)\nos.walk via _iter_candidate_files"]
    C --> E["_scan_scope(user)\nglob patterns via _iter_rule_candidate_files"]
    C --> F["_scan_scope(system)\nXDG_CONFIG_DIRS glob"]
    D & E & F --> G["_matching_rule()\nproduces ContextDiscoveryFinding per file"]
    G --> H["detect_target(project_root)"]
    H --> I["build_proposed_manifest()\nmerges existing apm.yml + findings"]
    I --> J["compute_migration_plan()\n_MIGRATION_MAP lookup (tool,kind) -> dest"]
    J --> K["ContextDiscoveryResult (frozen)"]
    K --> L["echo_discovery_result()\nclick.echo text/json/yaml"]
    L --> M{"write=True?"}
    M -- yes --> N["write_proposed_manifest()\ndump_yaml -> apm.yml"]
    N --> O["execute_migration(plan)\nshutil.copy2 / copytree per action"]
    M -- no --> P(["Preview shown; exit"])
    O --> Q(["apm.yml written + files migrated to .apm/"])

Loading

sequenceDiagram
    actor User
    participant CLI as init.py
    participant CD as context_discovery.py
    participant FS as Filesystem

    User->>CLI: apm init --discover --write
    CLI->>CD: discover_agent_context(cwd, config)
    CD->>FS: os.walk project tree
    FS-->>CD: file paths
    CD->>FS: glob user home patterns
    FS-->>CD: file paths
    CD-->>CLI: ContextDiscoveryResult (frozen)
    CLI->>User: echo preview (findings table + proposed apm.yml + migration plan)
    CLI->>User: Confirm? (y/N)
    User->>CLI: y
    CLI->>CD: write_proposed_manifest(result, apm.yml)
    CD->>FS: dump_yaml -> apm.yml
    CLI->>CD: execute_migration(plan)
    CD->>FS: shutil.copy2 / copytree per action
    FS-->>CLI: applied actions
    CLI->>User: Migration complete summary

Loading

Recommendation

Ship. The feature is architecturally sound, well-tested on its main paths, and delivers the highest-leverage adoption surface APM has shipped to date. No panelist found a blocking correctness regression or an actively exploitable security vulnerability that gates merge. Track five follow-ups as near-term issues: the README migration hook (same release window, not post-release), the two security findings (symlink and credential copying -- target before 1.0 or next minor), the --write->--apply rename (ship before this flag is widely adopted), and the two test-coverage gaps (compile pre-flight is the higher-urgency of the two). The doc-writer's --write description fix and importability definition are low-effort and can land in the same PR or a same-day follow-on commit. CLI logging polish (header, count divergence, cancellation symbol) should be bundled into a single cleanup commit in the same release.

---

Full per-persona findings

Python Architect

• [recommended] Triple-collision in compute_migration_plan() can produce two MigrationActions with the same dest at src/apm_cli/context_discovery.py:193
  When dest is already in seen_dests, the code reassigns dest to the {tool}-{dest_name} fallback and adds it to seen_dests. A third finding with the same (tool, dest_name) combination will again land on the same fallback path (already in seen_dests), skip the dest.exists() guard, and still append an action -- producing two actions targeting the same destination. execute_migration() will silently skip the second, but migration_plan in ContextDiscoveryResult will be misleading.
  Suggested: After the fallback reassignment, add: if dest in seen_dests: continue

• [recommended] USER_RULES rebuilds identical frozen dataclasses from TOOL_CONTEXT_RULES -- a pointless O(n) copy at src/apm_cli/context_discovery.py:665
  USER_RULES creates fresh ContextDiscoveryRule instances that are structurally identical to the originals. Simply alias: USER_RULES: tuple[ContextDiscoveryRule, ...] = TOOL_CONTEXT_RULES

• [nit] proposed_manifest in frozen dataclass is only shallowly frozen at src/apm_cli/context_discovery.py
  frozen=True prevents attribute reassignment but not dict mutation. Add a comment acknowledging the shallow freeze.

• [nit] write_proposed_manifest() uses a lazy intra-function import for dump_yaml while load_yaml and yaml_to_str are at module top at src/apm_cli/context_discovery.py
  Inconsistent import style. Promote to module-level.

• [nit] compile/cli.py: load_yaml as _load_yaml alias convention misused for a local binding at src/apm_cli/commands/compile/cli.py:405
  Use load_yaml (not _load_yaml) for a local variable binding -- the underscore prefix is a module-level lazy-import idiom.

• [nit] Design patterns are appropriate -- frozen-dataclass + collect-then-render + table-driven dispatch is correct at this scope

CLI Logging Expert

• [recommended] Header always reads 'Agent context discovery preview' even when --write is active at src/apm_cli/context_discovery.py
  During a write run that label is factually wrong. Branch on write param: write=True -> 'Agent context migration', write=False -> 'Agent context discovery preview'.

• [recommended] echo_discovery_result calls click.echo directly, bypassing CommandLogger quiet/verbose routing at src/apm_cli/context_discovery.py
  The raw click.echo will print the full multi-line report even under --quiet. Accept a quiet flag and skip text output when quiet=True.

• [recommended] Migration count reported twice on --write path and counts can silently diverge at src/apm_cli/commands/init.py
  echo_discovery_result(write=True) prints 'Migrating N files' before migration runs; then logger.success prints 'Migrated {len(applied)} files' after. When any dest exists, N != applied with no explanation.
  Suggested: Remove migration table from _format_text_result when write=True. Let logger.success carry the canonical count.

• [recommended] logger.progress('Initialization cancelled.') uses wrong symbol [>] for a terminal state at src/apm_cli/commands/init.py
  Use logger.info or logger.warning so the symbol accurately reflects a stopped state.

• [recommended] logger.progress('--yes specified, overwriting apm.yml...') leaks flag mechanics into human-facing output at src/apm_cli/commands/init.py
  Lead with outcome: logger.progress('Overwriting existing apm.yml...')

• [nit] No STATUS_SYMBOLS on any line in the text report at src/apm_cli/context_discovery.py
  Prefix informational lines with [i] and action-prompting lines with [*].

• [nit] 'Writing proposed apm.yml.' is present-tense but the write has not yet occurred at that point at src/apm_cli/context_discovery.py

DevX UX Expert

• [recommended] --write is a weak, misleading name for a flag that triggers file migration at src/apm_cli/commands/init.py
  Rename to --apply (terraform plan/apply mental model). Update error: '--apply can only be used with --discover.'

• [recommended] --format is silently accepted without --discover and silently ignored at src/apm_cli/commands/init.py
  Add guard: if output_format != 'text' and not discover: logger.error(...); sys.exit(1)

• [nit] Error message for non-existent project dir has a misleading recovery hint at src/apm_cli/commands/init.py
  'use --write to create a new one' implies --write bootstraps a project. Change to: '--discover scans an existing project. Run apm init my-project first, then re-run apm init --discover from inside it.'

• [nit] Docs migration end-to-end example second line has no inline comment at docs/src/content/docs/reference/cli-commands.md
  Add: # deploys .apm/ (now populated) to .codex/

Supply Chain Security Expert

• [recommended] shutil.copytree called without symlinks=True, silently dereferences symlinks inside copied skill directories at src/apm_cli/context_discovery.py
  An adversarial skill directory containing a symlink to /etc/passwd would have that file's contents written into .apm/. Pass symlinks=True to shutil.copytree or add a pre-copy symlink scan.

• [recommended] Token redaction is display-only; credential-bearing source files are copied verbatim to .apm/ at src/apm_cli/context_discovery.py
  If a scanned file contains a GitHub PAT or URL with embedded credentials, shutil.copy2 writes raw bytes into .apm/. Scan for token patterns before writing and warn-and-refuse or warn-and-redact.

• [nit] b'\x00' in chunk binary heuristic can misclassify UTF-16/UTF-32 text as binary at src/apm_cli/context_discovery.py
  Low security impact (files are skipped, not executed), but UTF-16 SKILL.md from Windows tools would be silently ignored.

• [nit] DEFAULT_SKIP_DIRS confirmed to include .git -- no path traversal risk from dest_name. Clean.

OSS Growth Hacker

• [blocking] README has zero mention of brownfield migration -- the Why do we need a GitHub token? #1 reason an existing Claude/Cursor/Codex user would install APM at README.md
  The two-command migration story does not appear anywhere in the README's first 100 lines. A visitor already using Claude Code will close the tab before reaching the CHANGELOG.
  Suggested: Add 'Already using Claude / Cursor / Copilot?' block immediately after the install snippet. Show the two-command migration and name the 7 tools it covers.

• [recommended] '--discover' is passive/technical; the user's mental model is 'migrate' at CHANGELOG.md
  Lead with the migration frame in release notes, docs, and social copy. Keep --discover as the flag but rename it conceptually in all user-facing copy.

• [recommended] The two-command migration sequence needs a dedicated quickstart section at docs/src/content/docs/getting-started/quick-start.md
  Feature has no discoverability story beyond word of mouth. Add a 'Brownfield migration' section lifting the worked example from CHANGELOG.

• [nit] 'Agent context discovery preview' as CLI output header is jargon; 'Migration plan (preview)' is clearer

• [nit] Tool list in README hero doesn't yet signal that APM can import FROM these tools, only deploy to them at README.md
  Change subtitle to: 'Install to any. Migrate from any.'

Doc Writer

• [recommended] --write option description understates its file-migration scope at docs/src/content/docs/reference/cli-commands.md
  The options list omits the primary action: copying convertible and misplaced files into .apm/. A brownfield user will be surprised by file movement.
  Suggested: '--write - With --discover, write apm.yml and migrate convertible/misplaced files into .apm/. Prompts for confirmation unless --yes is set.'

• [recommended] 'importability' is referenced in Behavior but never defined at docs/src/content/docs/reference/cli-commands.md
  Add inline definition: importability values: apm-native (already in .apm/ canonical form), convertible (can be migrated to .apm/), reference-only (detected but not migratable -- left in place).

• [recommended] Several CHANGELOG Fixed entries describe new behavior, not regressions at CHANGELOG.md
  'adds missing Codex/Copilot context rules' and 'wraps Claude plain .md skill files into SKILL.md directories' are net-new capabilities. Move those two entries to the Added section.

• [nit] --format default value not stated in options list at docs/src/content/docs/reference/cli-commands.md
  Change to: --format [text|json|yaml] - Discovery output format (default: text)

• [nit] Migration example second line has no inline comment at docs/src/content/docs/reference/cli-commands.md

• [nit] commands.md skill-file row description too verbose compared to other rows at packages/apm-guide/.apm/skills/apm-usage/commands.md
  Trim to match style of other entries.

Test Coverage Expert

• [recommended] _normalize_apply_to list-unwrap has no unit test -- silent drift risk on YAML applyTo lists at tests/unit/primitives/test_primitives.py
  Grep confirms zero hits for '_normalize_apply_to', 'applyTo.*[', or 'list.*applyTo' across all test files. The single-item unwrap and multi-item join paths are the user-visible contract.
  Proof (missing at unit): tests/unit/primitives/test_primitives.py::test_parse_instruction_list_apply_to -- proves: apm compile correctly handles .instructions.md files whose frontmatter has applyTo: ['**/*.py'], unwrapping single items and joining multi-item lists [portability-by-manifest, devx]

• [recommended] compile pre-flight includes-dir content probe has no test at any tier at tests/integration/
  Grep confirms zero hits for 'local_apm_has_content', 'includes.*dir', 'compile.*includes' across tests/. A regression would cause apm compile to abort for every project produced by the discover/write flow.
  Proof (missing at integration-with-fixtures): tests/integration/test_compile_preflight_includes_dir.py::test_compile_does_not_abort_when_content_in_includes_dir -- proves: apm compile does not abort when apm.yml lists an includes dir containing .instructions.md files (typical output of apm init --discover --write) [portability-by-manifest, devx]

Auth Expert -- inactive

No auth files changed (src/apm_cli/core/auth.py, token_manager.py, azure_cli.py, github_downloader.py, marketplace/client.py, github_host.py, install/validation.py, install/pipeline.py, deps/registry_proxy.py are all untouched); token redaction in context_discovery.py is output sanitization only and does not affect credential resolution or AuthResolver behavior.

_{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.

• #1227 pull_request_read: has lower integrity than agent requires. The agent cannot read data with integrity below "approved".
• feat(discovery): comprehensive brownfield discovery, migration, and cross-tool skill install (#1122) #1227 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 #1227 · ● 3.6M · ◷

[danielmeppiel]

@chkp-roniz Thanks for this! I am diving into the panel feedback and will come back ASAP. Direction is good