esonhugh/marketplace/open-case
Claude Code and agent skill by esonhugh/marketplace
- community
Checking that the SKILL.md file is still in the repository.
The assistant should use this when the user explicitly asks to open/create/start a Detective MCP case, begin an OODA investigation, or approves a brainstorm brief. It creates durable case state. Do not use for vague pre-case scoping without approval (use brainstorm), existing-case continuation (use investigate), status review (use review-board), blocker discussion (use discuss-case), or finalization (use close-case).
Install
npx vskill@latest install esonhugh/marketplace/open-caseSKILL.md
Open Case
Create a Detective MCP-backed case. Follow ../references/detective-core-protocol.md for canonical MCP-only state, enums, evidence lifecycle, SetGoal preference, StopHook fallback, and handoffs.
<HARD_GATE name="case-state-safety"> All case state creation and updates must go through Detective MCP tools. Never create, edit, rename, delete, or infer truth from .detective/ files as a substitute for MCP state. Preserve exact MCP tool names, enums, parameters, and signatures. </HARD_GATE>
<IMPORTANT name="goal-tool-boundary"> When a SetGoal or equivalent goal tool is available, use it only for harness/session orchestration after detective_open_case returns the case id. It does not create durable Detective case state and must not be treated as MCP truth. </IMPORTANT>
<HARD_GATE name="stop-fallback-marker-protocol"> Prefer SetGoal or an equivalent goal tool. Only when no goal tool is available, emit exactly once after detective_open_case returns: <DETECTIVE-STOP-FALLBACK status="active" case-id="<case_id>">. This only activates the statically registered inert StopHook fallback and must not be written to .detective/ files. </HARD_GATE>
Required Inputs
Ask one concise question at a time until you have title, description/crime scene, intent, budget, and scope.
Budget Mapping
Pass budget via detective_open_case(..., config={...}):
max_actionsor "N steps" →{"max_actions": N}.- Checkpoint cadence → add
"checkpoint_interval": <positive integer>. - Time box → add
"timebox": "<verbatim>". - If absent, use
{"max_actions": 8, "checkpoint_interval": 3}and state the default. - Seed coverage after opening with
detective_coverage_add; do not put coverage in config.
Exact Workflow
- Create durable case:
detective_open_case(title="<title>", description="<description>", case_id=null, config=<budget_config>)- Use
case_idonly if the user provided one; otherwise let MCP generate numbering/slug.
- Set orchestration only after durable case creation:
- If SetGoal or equivalent exists, set a concise goal with returned case id, intent, completion criteria, autonomous OODA instruction, and stop conditions.
- If unavailable, emit
<DETECTIVE-STOP-FALLBACK status="active" case-id="<case_id>">.
- Enter Observe:
detective_transition_phase(case_id="<case id>", phase="observe", reason="case opened")
- Record intent:
detective_add_intent(case_id="<case id>", intent="<user intent>", phase="observe", created_by="assistant", metadata={"source":"open-case"})
- Add an optional goal node only for graph reasoning or explicit success criteria:
detective_add_node(case_id="<case id>", type="question", content="Goal: <success criteria>", status="open", confidence=1.0, source="user", tags=["goal"], created_by="assistant", metadata={"is_goal": true})
- Seed graph from the brief:
- Crime scene →
type="observation",source="user",tags=["crime-scene"]. - Verified facts →
type="evidence"; partial signals →type="clue"; theories →type="hypothesis"; boundaries →type="constraint"; uncertainties →type="question".
- Link seed nodes where useful with
detective_add_edge(..., type="supports|contradicts|derives|requires|eliminates|related_to", ...). - Put uncertain scratch ideas on blackboard:
detective_blackboard_add(case_id="<case id>", content="<uncertain note>", kind="note", tags=["unverified"], created_by="assistant"). - Seed coverage with
detective_coverage_add(case_id="<case id>", area="<area>", status="planned", notes="seeded from open-case scope"); use"initial triage"if no areas are known. - Report case id, phase, intent, budget, seeded counts, coverage, goal-tool status, fallback status, and suggested
/detective:investigatenext step.
Output Format
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingplugins/detective-plugin/skills/open-case/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/esonhugh/marketplace/open-case)