Upload one skill to the Anthropic Skills API to validate the upload flow and confirm the skill loads correctly in a Managed Agent. Use the `skills/common/` skill since it's the simplest.
Scanned 5/29/2026
Install to Claude Code
npx -y skills add FDU-INS/Insurance-Skills --skill build-prompts --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Build Prompts?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/fdu-ins-build-prompts)More formats (shields.io, HTML) on the badges page.
# p0-05: Upload Test Skill to Skills API
## What to build
Upload one skill to the Anthropic Skills API to validate the upload flow and confirm the skill loads correctly in a Managed Agent. Use the `skills/common/` skill since it's the simplest.
## Context
The Anthropic Skills API allows uploading custom skills via `POST /v1/skills`. Skills require the `managed-agents-2026-04-01` beta header. The API returns a skill ID that you reference when creating an agent.
Read the Skills API documentation first. The Anthropic Python SDK should support skills — check via:
```bash
.venv/bin/python -c "import anthropic; help(anthropic.Anthropic().beta)"
```
If the SDK doesn't have direct skills support yet (it's beta), use raw HTTP requests via the `requests` library or `curl`.
## Tasks
1. **Read the current common skill** to understand what we're uploading:
```bash
cat skills/common/SKILL.md
ls skills/common/
```
2. **Create `deploy/push-skills.py`** — a minimal version that uploads a single skill. This will be expanded in Phase 2 to handle all skills from the manifest. For now, it should:
a. Load `ANTHROPIC_API_KEY` from `.env`
b. Read the skill file(s) from a given path
c. Upload via `POST /v1/skills` with the beta header
d. Print the returned skill ID
e. Save the skill ID to `deploy/skill-ids.json`
The script should accept a `--path` argument for the skill directory and a `--name` argument for the skill identifier.
Usage:
```bash
.venv/bin/python deploy/push-skills.py \
--path ./skills/common/ \
--name skill_healthsim_common_v1
```
3. **Run the upload**:
```bash
source .env
.venv/bin/python deploy/push-skills.py \
--path ./skills/common/ \
--name skill_healthsim_common_v1
```
4. **Verify the upload** by listing skills:
```bash
curl -s https://api.anthropic.com/v1/skills \
-H "x-api-key: $ANTHROPIC_API_KEY" \
-H "anthropic-version: 2023-06-01" \
-H "anthropic-beta: managed-agents-2026-04-01" | python -m json.tool
```
5. **Record the skill ID** in `deploy/skill-ids.json`:
```json
{
"skill_healthsim_common_v1": "skill_01abc..."
}
```
6. **Git commit**: "feat: add push-skills.py and upload test skill"
## Deliverables
- `deploy/push-skills.py` (minimal, single-skill version)
- `deploy/skill-ids.json` with the uploaded skill ID
- Verified: skill appears in Skills API listing
- Committed
## Important
- The Skills API is in beta. If the endpoint structure or request format differs from what's documented here, adapt accordingly. The Anthropic docs at `platform.claude.com/docs/en/agents-and-tools/agent-skills/` are the authoritative source.
- If multi-file skills (SKILL.md + supporting .md files) need special handling for the API, document what you learn — we'll need this knowledge for Phase 2 when uploading all 9 skills.
- Do NOT modify the skill content yet. Upload it as-is. Adaptation (replacing embedded data refs, adding output format sections) happens in Phase 2.
No comments yet. Be the first to comment!