Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/francis-du/wcode/wcode-software-intelligencenpx skills add francis-du/wcode --skill wcode-software-intelligencegit clone --depth 1 https://github.com/francis-du/wcodeWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00031 | $0.00666 |
| Opus 5 | $0.00015 | $0.00333 |
| Sonnet 5 | $0.00006 | $0.00133 |
| Haiku 4.5 | $0.00003 | $0.00067 |
Grade A, and why
wcode-software-intelligence scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
Use the configured wcode MCP server for repository work. Keep instructions short and use progressive disclosure: load only the project, source, design, or verification detail needed for the current task.
Before editing:
- Call
workspace_infoand use the closest discovered subspace for the task. - Start with
agent_context(goal, scopes=...). Callscope_status,design_status,project_context, orlanguage_quality_statusonly when readiness or the task requires deeper inspection. - Prefer
find_symbol,file_outline,search_code, andsymbol_contextfor cheap localization. Usesemantic_navigationonly when readiness or the task needs cross-file references, callers/callees, implementations, hover, or semantic impact; preferpath + symbolover model-computed LSP offsets. - Treat Tree-sitter facts as
precision=syntax; only fresh provider facts are semantic. Hardened first-party semantics may be maintained automatically;--no-semanticis the operator opt-out.
While editing:
- Stay inside the selected Workspace. Preserve SHA-256 preconditions and use bounded wcode edit tools.
- When targets are known, prefer one
apply_edits,apply_file_edits, orcreate_filescall. Parallelize only independent work. - Reuse the canonical model or helper before adding another wrapper, mode, or special case. A file crossing 1,000 lines needs decomposition or an explicit structural reason.
- Treat
delete_pathas exceptional. Never bypass path, symlink, hard-link, no-shell, or command-policy checks. - Authorization has separate scopes:
CommandAccessgrants one executable in one Workspace;RiskyExecutiongrants only the fingerprinted trust requested by the operation. That is normally one exact repository operation, while a non-automatic language server uses Workspace + Provider + current binary identity trust so a warm session can be reused without silently widening to a replacement binary or other providers. Surface the request, let the operator decide in the TUI or protected WebUI, then retry only the matching operation. - Never enable
--allow-risky-exec, approve a request, or widen a Workspace on the user's behalf.
After editing:
- Run
review_changes; treatmaintainability-*findings as structural signals. - Use
drift_status,impact_analysis, andrisk_statuswhen the change needs them. - Run the recommended
verify_projectlevel. Uselanguage_quality_runonly for a declared, available, check-only provider. - Continue
reconciliation_planwhen traceability or drift gaps remain. - Finish with
evidence_status; report failures, disagreement, stale revisions, and remaining blockers. Never invent a Stage Pass or HumanApproval.
One producer's Pass does not erase another producer's latest Fail.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 60 lines · 31 tokens per session scan A d4e3c62aaa1c
wcode-software-intelligence is a skill published in the GitHub repository francis-du/wcode (5 stars, last pushed 2d ago), licensed Apache-2.0. It adds 31 tokens to every session and 666 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
code_interpreter
在隔离沙盒中执行 Python 代码，用于数据分析、数值计算、格式转换、图表数据生成等需要真正运行代码才能得出结果的任务。.
code-review
Comprehensive code review for security, performance, and best practices.
smoke-test
End-to-end smoke test skill for DeerFlow. Guides through: 1) Pulling latest code, 2) Docker OR Local installation and deployment (user preference, default to Local if Docker network issues), 3) Service availability verification, 4) Health check, 5) Final test report. Use when the user says "run smoke test", "smoke…
image-generation
Use this skill when the user requests to generate, create, imagine, or visualize images including characters, scenes, products, or any visual content. Supports structured prompts and reference images for guided generation.
podcast-generation
Use this skill when the user requests to generate, create, or produce podcasts from text content. Converts written content into a two-host conversational podcast audio format with natural dialogue.
vercel-deploy
Deploy applications and websites to Vercel. Use this skill when the user requests deployment actions such as "Deploy my app", "Deploy this to production", "Create a preview deployment", "Deploy and give me the link", or "Push this live". No authentication required - returns preview URL and claimable deployment link.