Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

RDC Debugger 主技能包装说明 — measured AI agent skill | DecimalAI

haolange/rdc-debugger

SkillScore91Maintained

Public main skill for the RenderDoc/RDC GPU debugger framework. Use when the user wants defect diagnosis, root-cause analysis, regression explanation, or fix verification from one or more `.rdc` captures. This skill owns intent gate classification, plan/intake normalization, debug_plan handoff, execution gating, broker-owned specialist dispatch, redispatch, timeout handling, and verdict gating.

by haolange

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Coding pack

decimalai skills pull haolange-rdc-debugger

Get this skill — curl · raw file

curl -fsSL https://app.decimal.ai/s/haolange-rdc-debugger/SKILL.md

Raw file: /s/haolange-rdc-debugger/SKILL.md ↗ · machine-readable summary at /s/haolange-rdc-debugger.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 6d ago

Yes — +64% lift+3% tokens0% turns91% of whole cases pass with skill (20/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-03	✗→✓	▲ Improved	-12%	0%
case-01	✗→✓	▲ Improved	-50%	0%
case-05	✗→✓	▲ Improved	8%	0%
case-06	✗→✓	▲ Improved	25%	0%
case-07	✗→✓	▲ Improved	55%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/1/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

haolange / rdc-agentcaps

debugger/common/skills/rdc-debugger/SKILL.md

View on GitHub ↗

License NOASSERTIONImported 8/1/2026Version v6

Skill contents

SKILL.md

`RDC Debugger` 主技能包装说明

当前文件是 Codex 的 public main skill 入口。

平台启动后默认保持普通对话态。只有用户手动召唤 rdc-debugger，才进入 RenderDoc/RDC GPU Debug 调试框架。

rdc-debugger 是单一公开入口，但内部固定拆成两段：

Plan / Intake Phase
Audited Execution Phase

Plan / Intake Phase

当前阶段兼容宿主 Plan Mode，负责：

intent_gate
缺失输入补料
reference_contract 生成与 readiness 判定
debug_plan 编译

当前阶段对外只允许要求用户补充自然语言事实、目标、现象、环境和可访问材料。

不得把以下内部对象直接当成用户输入要求：

reference_contract
strict_ready
missing_inputs
debug_plan

如果 execution 仍被 BLOCKED_MISSING_FIX_REFERENCE 阻断，必须明确说明缺的是哪类参考事实或基线材料，而不是让用户按内部字段名补 YAML。

当前阶段默认通过轻量 sub-agent 收敛输入：

clarification_agent
reference_contract_agent
plan_compiler_agent

当前阶段不允许：

创建 case/run
写入 run/session 审计产物
进入 specialist dispatch 或 live rd.* 分析

Audited Execution Phase

只有当 debug_plan.execution_readiness = ready 时，当前 skill 才允许进入 execution。

固定顺序：

entry_gate
accept_intake，内部完成 capture import、case/run bootstrap、intake_gate 与 broker startup
triage
dispatch/specialist loop
skeptic
curator
final_audit

在 artifacts/intake_gate.yaml 通过，且 artifacts/runtime_session.yaml、artifacts/runtime_snapshot.yaml、artifacts/ownership_lease.yaml、artifacts/runtime_failure.yaml 产出前，不得进入 specialist dispatch 或 live rd.* 分析。

本 skill 只引用当前平台根目录的 common/：

common/skills/rdc-debugger/SKILL.md
进入任何平台真相相关工作前，必须先校验 common/config/platform_adapter.json
local_support / remote_support / enforcement_layer / coordination_mode 统一以 common/config/platform_capabilities.json 的当前平台定义为准

当前平台固定 contract：

coordination_mode = staged_handoff
orchestration_mode = multi_agent
live_runtime_policy = single_runtime_single_context
hook_ssot = shared_harness

执行约束：

live tools process 始终由 broker / coordinator 直接持有。
specialist 只能在有效 ownership_lease 下通过 broker action 请求访问 live runtime。
不得直接持有 live CLI / process，不得缓存并跨 handoff 复用 runtime handle。
不得临时写 Python、PowerShell 或 shell wrapper 批处理 live CLI。
triage + specialist + skeptic + curator 明确必须通过 sub-agent 执行。
entry_gate、accept_intake、intake_gate、dispatch_readiness、final_audit 属于控制面，不 agent 化。
specialist dispatch 后，主 agent 必须进入 waiting_for_specialist_brief 并持续汇总阶段回报；短时 silence 不得触发 orchestrator 抢活。
超过框架预算仍未收到阶段回报时，必须进入 BLOCKED_SPECIALIST_FEEDBACK_TIMEOUT 或等价阻断状态，而不是让 orchestrator 抢做 specialist live investigation。

未先将顶层 debugger/common/ 拷入当前平台根目录的 common/ 之前，不允许在宿主中使用当前平台模板。

运行时 case/run 现场与第二层报告统一写入平台根目录下的 workspace/。

Want to see if haolange/rdc-debugger would activate on your phrasing? Try it before you install — no LLM call, no signup.

Other measured skills in the registry, with their headline benchmark lift.