菜单
Check upstream tool documentation for changes that affect docs/cross-tool-strategy.md. Fetches source URLs, compares claims against current doc, and reports drift. Use when updating the cross-tool strategy guide or checking for tool changes.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | research-update |
| description | Check upstream tool documentation for changes that affect docs/cross-tool-strategy.md. Fetches source URLs, compares claims against current doc, and reports drift. Use when updating the cross-tool strategy guide or checking for tool changes. |
| compatibility | ["claude-code"] |
| metadata | {"version":"1.0","author":"team"} |
| Scope | What It Checks |
|---|---|
| (all) | All sources |
claude-code | Claude Code docs only |
copilot | GitHub Copilot CLI docs only |
opencode | OpenCode docs only |
community | Community resources only |
Fetch each URL listed in docs/cross-tool-strategy.md Section 4 (Sources), filtered by the selected scope. That section is the single authoritative source list — do not invent URLs, and do not maintain a copy here. A URL added to or retired from the doc is automatically in or out of scope.
Fetch each URL in the selected scope. If a URL returns an error (404, 403, timeout), note it as a potential doc restructure.
For each source, extract the current state of:
| Category | What to Look For |
|---|---|
| Frontmatter fields | New, renamed, or deprecated YAML fields for agents/skills |
| Discovery paths | Changes to where tools look for agents, skills, rules files |
| Tool permissions | New tool names, permission syntax changes |
| Model support | New models, changed model IDs, deprecated models |
| Multi-agent | Agent Teams status changes, new orchestration features |
| Compatibility | Changes to cross-tool behavior (CLAUDE.md reading, AGENTS.md fallback) |
| New features | Capabilities not mentioned in the current doc |
| Deprecations | Features marked deprecated or removed |
Read docs/cross-tool-strategy.md and compare each claim against what the sources say now. Also read the version-stamped surfaces of docs/specialist-agent-workflow.md: the §1 Agent Teams status and cost claims, the §4 frontmatter and model-pin table, and the §6 migration playbook's install steps. Flag:
## Research Update: [date]
### Scope: [all | claude-code | copilot | opencode | community]
### Outdated Claims
1. **Section N.N [title]** — line NN
- **Current doc:** [what it says]
- **Source says:** [what the source now says]
- **Source URL:** [url]
- **Suggested fix:** [specific edit]
### New Features (not in doc)
1. **[tool] [feature name]**
- **What:** [description]
- **Source URL:** [url]
- **Where to add:** Section N.N
- **Impact on projects:** [does this affect samples/go or samples/java-spring-boot configs?]
### Deprecations
1. **Section N.N [title]** — line NN
- **What:** [feature] is now deprecated
- **Source URL:** [url]
- **Suggested fix:** [specific edit]
### Broken URLs
1. **Section N.N** — line NN
- **URL:** [url]
- **Status:** [404 | redirect to X | timeout]
### Version Drift
1. **Section N.N** — line NN
- **Current doc:** [old version/date]
- **Current:** [new version/date]
### No Changes
- [list sections that are still accurate]
### Summary
- X outdated claims
- Y new features to document
- Z deprecations
- W broken URLs
- V version updates needed
For each finding, propose a specific edit to docs/cross-tool-strategy.md (or to the named specialist-agent-workflow.md surfaces when those drift):
Do NOT apply edits automatically. Present them for review.
After edits are applied, recommend:
harness/verify-harness.py --quick (tier 0 for a docs-only edit); before committing a substantive change, /audit-harness per the maintainer loop in the root CLAUDE.md.Status: snapshot date at the top of docs/cross-tool-strategy.md.If two sources disagree:
TDD cycle process and design-check decision tree for feature implementation. Load when implementing features using test-driven development.
Install or update the claude-pod tooling — the claude-pod command, the pod Dockerfile, the default config, the IDE-oracle preflight, and the egress filter pair — from this repo's tools/claude-pod/ into ~/.local/bin and ~/.config/claude-pod. Thin front-end for tools/claude-pod/install.sh: run its check mode to show drift, apply only on the user's approval. Use when the user asks to install claude-pod, set up the container pod, or update the pod script or image definition. Examples: "install claude-pod", "update the pod tooling", "set up the claude container".
Install or update the harness-stats tooling — harness-statusline.sh, cache-report.sh, and the cache-report skill — from this repo's tools/harness-stats/ into the user's ~/.claude/ directory. Thin front-end for tools/harness-stats/install.sh: run its check mode to show drift, apply only on the user's approval. Use when the user asks to install the cache tooling, set up the statusline, update the harness-stats scripts, or reconfigure cache reporting. Examples: "set up the statusline", "install harness-stats", "update the cache tooling".
Generate a per-agent cache-efficiency report for the current Claude Code session. Use when the user asks about cache stats, cache efficiency, prompt-cache performance, token savings, or whether subagents are using cache effectively. Examples: "show cache report", "how's the cache doing", "cache stats", "are subagents efficient", "which agents are wasting cache", "session token breakdown".
Periodic multi-angle improvement review of the reference. Always fans out all five read-only research agents in parallel — maintainer tooling, documentation, runtime cost, source duplication, consumer surface — so cross-angle tension informs the synthesis. Every agent carries the challenge charter: question the design against the README's stated philosophy, recommend removals that pay for themselves. Judges findings by the resilience-first doctrine; a settled ADR decision is rebuttable with a new argument, never silently suppressed. "/review-harness challenge" drops the ADR anchoring from the agent prompts for a zero-based outside look. Adversarially verifies structural findings and delivers one prioritized report. Assessment only — never edits. Complements /audit-harness, which gates defects against the current bar. Root-only (Claude Code).
Build, test, format, and lint requirements that must pass before code review. Load when checking implementation completeness or running the quality gate.