What needs to be fixed?
Context
PR #171 / #165 sets manifest.yaml name to finance/wallet_screening (registry ID). That is correct, but several runnable examples and docs still assume the legacy short tool name wallet_screening.
SkillLoader adapters derive tool names from manifest["name"]:
| Adapter |
Tool name after rename |
| Gemini / Claude |
finance/wallet_screening |
| OpenAI / DeepSeek |
finance_wallet_screening (sanitized) |
| Ollama prompt |
finance/wallet_screening |
Runnable scripts that hardcode wallet_screening will not dispatch tool calls after #171 merges.
What to update
Examples (runtime fix — prefer bundle["manifest"]["name"] like claude_evm_tx_handler.py / gemini_evm_tx_handler.py):
examples/gemini_wallet_check.py — if fn_name == "wallet_screening"
examples/claude_wallet_check.py — if tool_name == "wallet_screening"
Skill cognition:
skills/finance/wallet_screening/instructions.md — still says `wallet_screening`
Docs (tool-name tables and comments):
docs/skills/wallet_screening.md — Claude/Gemini/OpenAI/Ollama sections reference wallet_screening
docs/usage/openai.md — table row wallet_screening → wallet_screening
docs/introduction.md — example mentions wallet_screening tool
Optional consistency (lower priority):
COMPARISON.md — prose examples using short name (ID vs tool name is fine to clarify)
Out of scope
- Loader / adapter behavior (working as designed)
- Other skills with legacy short manifest names (
pdf_form_filler, etc.) — separate normalization issue if desired
Acceptance criteria
Related
Affected Page
examples/gemini_wallet_check.py, examples/claude_wallet_check.py, skills/finance/wallet_screening/instructions.md, docs/skills/wallet_screening.md, docs/usage/openai.md, docs/introduction.md
What needs to be fixed?
Context
PR #171 / #165 sets
manifest.yamlnametofinance/wallet_screening(registry ID). That is correct, but several runnable examples and docs still assume the legacy short tool namewallet_screening.SkillLoaderadapters derive tool names frommanifest["name"]:finance/wallet_screeningfinance_wallet_screening(sanitized)finance/wallet_screeningRunnable scripts that hardcode
wallet_screeningwill not dispatch tool calls after #171 merges.What to update
Examples (runtime fix — prefer
bundle["manifest"]["name"]likeclaude_evm_tx_handler.py/gemini_evm_tx_handler.py):examples/gemini_wallet_check.py—if fn_name == "wallet_screening"examples/claude_wallet_check.py—if tool_name == "wallet_screening"Skill cognition:
skills/finance/wallet_screening/instructions.md— still says`wallet_screening`Docs (tool-name tables and comments):
docs/skills/wallet_screening.md— Claude/Gemini/OpenAI/Ollama sections referencewallet_screeningdocs/usage/openai.md— table rowwallet_screening→wallet_screeningdocs/introduction.md— example mentionswallet_screeningtoolOptional consistency (lower priority):
COMPARISON.md— prose examples using short name (ID vs tool name is fine to clarify)Out of scope
pdf_form_filler, etc.) — separate normalization issue if desiredAcceptance criteria
gemini_wallet_check.pyandclaude_wallet_check.pydispatch using manifest-derived name (or updated literal matching adapter output)instructions.mdtells the agent the correct tool name from manifestdocs/skills/wallet_screening.mdanddocs/usage/openai.mddocument correct per-provider tool namesexamples/README.mdCHANGELOG.md[Unreleased]entry if user-visible docs/examples changeRelated
Affected Page
examples/gemini_wallet_check.py, examples/claude_wallet_check.py, skills/finance/wallet_screening/instructions.md, docs/skills/wallet_screening.md, docs/usage/openai.md, docs/introduction.md