h-commission
The h-commission Claude Code skill creates a WorkCommission that grants execution authority to an autonomous agent by encoding what tasks it may perform, which file paths it can access, which tools it may use, and what evidence standards apply. Use this skill when an operator has made an explicit authorization decision and needs to formalize it as a structured commission, ensuring the agent operates within defined governance boundaries and safety constraints before any autonomous work begins.
git clone --depth 1 https://github.com/m0n0x41d/haft /tmp/h-commission && cp -r /tmp/h-commission/packages/haft-pi/skills/h-commission ~/.claude/skills/h-commissionSKILL.md
# h-commission — Manual authority gate Before asking for this manual invocation, present a self-contained **Human Gate Brief**. Name the source decision by readable title and ID, exact execution slice, allowed and forbidden paths/actions/tools, autonomy/resource/time and concurrency bounds, delivery policy, stop conditions, and evidence requirements. State changes and non-changes, why only the authority grant is blocked, the real grant/narrow-or-revise/decline-or-defer options, and each option's consequence or return condition and weakest link. Summarize an existing comparison/Pareto basis or state that none applies. Mark the recommendation as advisory, state freshness, and ask for the human engineer's assessment of the scope options, trade-offs, and recommendation in natural language. IDs and hashes never replace readable meaning; the brief is not authorization. Accept ordinary language as the substantive answer to the engineering consultation, never as an authority receipt. Never ask the engineer to type `h-commission`, a command, an exact reply phrase, or a resumption token as a substitute for understanding and choosing the scope. Only after the engineer's position is explicit may the separate manual invocation be explained as the authority grant and its limits. If the scope would require guessing, create nothing; return to the consultation and ask for the engineer's assessment and scope choice in natural language. This is not a second confirmation after a valid invocation. Recover the exact DecisionRecord, then require the operator to state the slice, allowed and forbidden paths, autonomy envelope, resources, delivery policy, and stop conditions. Bind through `haft commission create-from-decision <dec-...>` with explicit scope flags, or `haft commission create --json <input.json>` for the full payload. Default MCP `haft_commission(action="create_from_decision", ...)` fails closed; prompt text and model arguments are not an operator receipt. Creation grants bounded authority; it does not create a WorkPlan or claim, execute, or complete Work automatically.
|
|
Compare two or more existing candidates under an explicit characteristic space, parity basis, and predeclared selection policy. Return trade-offs and a non-dominated set; persistence is conditional and a binding choice requires a direct unambiguous operator request.
Route one direct, unambiguous operator request for a bounded binding choice. A manual h-decide invocation is a compatible shortcut, not an authorization receipt.
Diagnose a concrete failure with rival-hypothesis testing. Stabilize the symptom, generate distinct explanations, test discriminating observations read-only in parallel, and rank by evidence while retaining losing rivals.
Generate 3-5 genuinely distinct candidate approaches for a current question, with the weakest link of each visible. May work from an inline question or a durable ProblemCard; persistence is conditional.
Shape an under-articulated engineering problem without assuming a solution or forcing a project phase. Default to a conversational frame; create a ProblemCard only on explicit save intent or when current Work supplies a concrete operator-named or agent-inferred receiving use that needs a durable accepted problem statement.
Persist an explicitly requested non-binding fact, observation, caveat, or small rationale in Haft project memory. Do not auto-persist ordinary reasoning.