[BUG] apm fails to install or compile dependencies

[petemounce]

Describe the bug

apm fails at core functionality.

To Reproduce
Steps to reproduce the behavior:

1. Make an apm.yml with 1 dependency:

   name: x
   version: 0.1.0
   description: APM project for x
   author: Peter Mounce
   dependencies:
     apm:
     - JuliusBrussee/caveman#84cc3c14
     mcp: []
   scripts: {}
   target: opencode,claude,copilot,agents

2. Run apm install

3. Command completes. apm_modules/ is created with the dependency within.

4. No content is "deployed" to the filesystem of the repository. .claude/skills, .github/skills, .opencode/skills, .agents/skills are all empty.

5. Run apm compile

6. No content is "deployed" to the filesystem of the repository. .claude/skills, .github/skills, .opencode/skills, .agents/skills are all empty.

Expected behavior

I expect the caveman skills to be deployed into the places where the agents will load them, based on reading the documentation's quick-start.

Instead, I get successful exit codes and log messages - but no actions appear to have been actually taken.

Environment (please complete the following information):

• OS: macOs
• Python Version: 3.14.3
• APM Version: 0.8.12
• VSCode Version (if relevant): n/a

Logs

apm install --verbose
[>] Installing dependencies from apm.yml...
Parsed apm.yml: 1 APM deps, 0 MCP deps
Using apm.lock.yaml (1 locked dependencies)
    JuliusBrussee/caveman: locked at 84cc3c14
Resolved 1 direct dependencies (no transitive)
  [+] JuliusBrussee/caveman #84cc3c14 @84cc3c14 (cached)

No MCP dependencies found in apm.yml

[*] Installed 1 APM dependency.

apm compile --clean
[*] Starting context compilation...
[i] Compiling for AGENTS.md + CLAUDE.md + .github/ + .claude/ + .cursor/ + .opencode/ + .codex/ + .agents/ - detected .github/ and .claude/ and .opencode/ folders
[+] Compilation completed successfully!

ls -al .opencode .claude .github
.claude:
total 16
drwxr-xr-x@  4 pmounce  staff   128 21 Apr 17:32 .
drwxr-xr-x@ 45 pmounce  staff  1440 21 Apr 18:12 ..
-rw-r--r--@  1 pmounce  staff    32 16 Apr 11:46 .gitignore
-rw-r--r--   1 pmounce  staff    18 21 Apr 17:36 settings.json

.github:
total 24
drwxr-xr-x@  9 pmounce  staff   288 21 Apr 17:32 .
drwxr-xr-x@ 45 pmounce  staff  1440 21 Apr 18:12 ..
drwxr-xr-x@  7 pmounce  staff   224 16 Apr 11:46 actions
-rw-r--r--@  1 pmounce  staff  1365 16 Apr 11:46 CODEOWNERS
-rw-r--r--@  1 pmounce  staff  2006 16 Apr 11:46 dependabot.yml
drwxr-xr-x   3 pmounce  staff    96 21 Apr 17:32 hooks
-rw-r--r--@  1 pmounce  staff     0 16 Apr 11:46 pull_request_template.md
drwxr-xr-x@  5 pmounce  staff   160 16 Apr 11:46 workflows
-rw-r--r--@  1 pmounce  staff    54 16 Apr 11:46 zizmor.yml

.opencode:
total 0
drwxr-xr-x   2 pmounce  staff    64 21 Apr 18:03 .
drwxr-xr-x@ 45 pmounce  staff  1440 21 Apr 18:12 ..

tree apm_modules/JuliusBrussee/caveman
apm_modules/JuliusBrussee/caveman
├── AGENTS.md
├── benchmarks
│   ├── prompts.json
│   ├── requirements.txt
│   ├── results
│   └── run.py
├── caveman
│   └── SKILL.md
├── caveman-compress
│   ├── README.md
│   ├── scripts
│   │   ├── __init__.py
│   │   ├── __main__.py
│   │   ├── benchmark.py
│   │   ├── cli.py
│   │   ├── compress.py
│   │   ├── detect.py
│   │   └── validate.py
│   ├── SECURITY.md
│   └── SKILL.md
├── caveman.skill
├── CLAUDE.md
├── CLAUDE.original.md
├── commands
│   ├── caveman-commit.toml
│   ├── caveman-review.toml
│   └── caveman.toml
├── CONTRIBUTING.md
├── docs
│   └── index.html
├── evals
│   ├── llm_run.py
│   ├── measure.py
│   ├── plot.py
│   ├── prompts
│   │   └── en.txt
│   ├── README.md
│   └── snapshots
│       └── results.json
├── gemini-extension.json
├── GEMINI.md
├── hooks
│   ├── caveman-activate.js
│   ├── caveman-config.js
│   ├── caveman-mode-tracker.js
│   ├── caveman-statusline.ps1
│   ├── caveman-statusline.sh
│   ├── install.ps1
│   ├── install.sh
│   ├── package.json
│   ├── README.md
│   ├── uninstall.ps1
│   └── uninstall.sh
├── LICENSE
├── plugins
│   └── caveman
│       ├── assets
│       │   ├── caveman-small.svg
│       │   └── caveman.svg
│       └── skills
│           ├── caveman
│           │   ├── agents
│           │   │   └── openai.yaml
│           │   ├── assets
│           │   │   ├── caveman-small.svg
│           │   │   └── caveman.svg
│           │   └── SKILL.md
│           └── compress
│               ├── scripts
│               │   ├── __init__.py
│               │   ├── __main__.py
│               │   ├── benchmark.py
│               │   ├── cli.py
│               │   ├── compress.py
│               │   ├── detect.py
│               │   └── validate.py
│               └── SKILL.md
├── README.md
├── rules
│   └── caveman-activate.md
├── skills
│   ├── caveman
│   │   └── SKILL.md
│   ├── caveman-commit
│   │   └── SKILL.md
│   ├── caveman-help
│   │   └── SKILL.md
│   ├── caveman-review
│   │   └── SKILL.md
│   └── compress
│       ├── scripts
│       │   ├── __init__.py
│       │   ├── __main__.py
│       │   ├── benchmark.py
│       │   ├── cli.py
│       │   ├── compress.py
│       │   ├── detect.py
│       │   └── validate.py
│       └── SKILL.md
└── tests
    ├── caveman-compress
    │   ├── claude-md-preferences.md
    │   ├── claude-md-preferences.original.md
    │   ├── claude-md-project.md
    │   ├── claude-md-project.original.md
    │   ├── mixed-with-code.md
    │   ├── mixed-with-code.original.md
    │   ├── project-notes.md
    │   ├── project-notes.original.md
    │   ├── todo-list.md
    │   └── todo-list.original.md
    ├── test_hooks.py
    └── verify_repo.py

31 directories, 83 files

Additional context
I have experimented with adding path: skills/caveman to the dep; that changes the log output such that a skill is recognised - but still no content is placed into my git working tree.

[danielmeppiel]

@petemounce the target logic in the manifest has been updated and the docs reflect stuff in HEAD instead of v0.8.12. I'll ship v0.9.0 and we will work on fixing the docs issue. As a workaround, ensure you create the .github and related folders and they should be populated. If they do not exist, v0.8.12 won't create them.

[danielmeppiel]

@petemounce can you try with v0.9.0 that was just released?