Fulcrum 10 of 18
Handing work to a subagent
When an agent delegates a piece of work to another agent (a subagent), and when that subagent reports back.
What goes wrong here
The subagent gets a compressed prompt that says what to do but not why. It fills the gap with its own interpretation, builds to that, and reports 'done' based on its own assessment. The parent passes the report along.
How it compounds if nothing catches it
Delegation multiplies drift. Each handoff is another translation, and a fleet of subagents can produce a lot of confident work in parallel against a misread target. Reports that subagents grade themselves pass the misreading up the chain as if it had been verified.
What the harness does here
When a subagent starts, a hook scores its task. It also requires the subagent to state the target experience ('The person using this will see…') before writing code, and to search institutional memory first. When the subagent stops, the evidence gate that guards the main agent checks its completion claims too. A second hook logs whether it stated the target experience. A third saves any research reports it produced so they aren't lost with its context. execute2 is the dispatch pattern: read the parent context, carry a verification contract that could fail, and return proof the parent can re-run. In orchestrator and execute, the work is checked by an evaluator that never saw the builder's reasoning, on the premise that agents grading their own work tend to praise it.
The pieces that act here
- subagent-start-governer.sh
A hook: a script Claude Code runs on its own at a set moment.
- subagent-stop-alignment-check.sh
A hook: a script Claude Code runs on its own at a set moment.
- stop-evidence-gate.sh
A hook: a script Claude Code runs on its own at a set moment.
- hook-extract-on-stop.sh
A hook: a script Claude Code runs on its own at a set moment.
- /execute2
Dispatch a single subagent on a task with full alignment discipline baked in — ascendant-context read, falsifiable verification contract, RECOVERABLE_DRAFTS handoff, and parent self-enforcing loop. Use when you (parent agent) want to delegate a discrete task to a subagent and get back work that you can independently verify with runnable commands instead of trusting self-assessment. Portable: works for any task in any project. Validated end-to-end in a real project (one dispatch, 9 of 9 checks passed); this skill extracts that pattern as a reusable command.
- /execute
Adversarial execution pipeline for a confirmed task list. Uses a session-persistent team (scope guardian, compliance auditor) when your Claude Code environment offers the experimental multi-agent team tools; otherwise runs the same guardian/auditor jobs as ordinary per-batch subagents with their findings carried forward in a local log file. Dispatches implementers with independent cross-verification, adversarial disproof, and a response-predictor gate. Use when you have a confirmed task list and need verified implementation.
- /orchestrator
Use when a top-level agent must define outcomes, route work across analysis, research, and implementation layers, preserve evidence and accountability across every hand-off, and keep the person it's working with immediately informed without ever waiting on subagent work.
- /nested-intent2
Produce a nested intent map for a multi-layered initiative — top-level goal, child goals, sub-goals, cross-cutting intents — where each node carries an intent paragraph, in the voice of the person you're working with if you've calibrated it, with enough precision that a subagent reading just that node can act at 100% fidelity. Use when the human asks for nested intent on an initiative with multiple sub-goals, when a plan needs to be turned into a context-managed handoff package for subagents, or when an existing outline needs intent paragraphs added at each layer of nesting.
- /dispatch-seed
Pipeline from intent alignment to agent dispatch. Runs /align until confirmed, creates an observable seed file, generates a plan with a docs-first approach, then dispatches. The full lifecycle from "I want this done" to "an agent is working on it.