| name | chat2api-tool-client-replay |
| description | Use when replaying sanitized client tool-calling fixtures against Chat2API with client-specific pass/fail rules. |
Chat2API Tool Client Replay
Use this skill to replay fixture scenarios for clients such as Cherry Studio, generic OpenAI tools, or unknown HAR-derived clients.
Rules
- Preserve exact tool names.
- Validate stream and non-stream behavior separately.
- Use profiles for client-specific prompt-protocol expectations.
Commands
CHAT2API_API_KEY=sk_xxx \
node skills/chat2api-tool-client-replay/scripts/replay-client-fixture.mjs \
--fixture backup/har/cherry-studio-fixture.json \
--profile cherry-studio \
--model deepseek-v4-flash
Use --dry-run to check profile, model, and fixture selection without live requests. If the fixture file exists, dry-run parses the fixture schema and reports fixtureExists plus scenarioCount; if it is missing, dry-run still exits successfully with fixtureExists: false.
Profiles
cherry-studio: Cherry Studio headers and visible prompt-tool-use expectations.openai-tools: generic OpenAI-compatible tool-calling expectations.custom-har: empty client headers for replaying sanitized HAR-derived fixtures.
