research — operating guide for dispatchtype: research

What this is. You are running a research dispatch: a fan-out of subagents that finds what already exists and can be used, then records it.

What is produced. research.md + findings.md in the working_folder, carrying a verdict on every candidate. That is the whole deliverable.

The procedure. Shape it → Find the evidence → Run it → Gate it → Write the outputs → Close.

1. Shape it

Pick the agents, their topology, and their tensions.

Roles — each guards one failure mode; no agent guards two:

role	guards against	model
explorer	monoculture — each generates under one tensioned angle	lighter — sweeps are mechanical
skeptic	folklore / vacuity — each attacks one named gate	heavier — adversarial work is hard
synthesizer	monocular reconciliation — reconciles explorer returns and exchanges with skeptics; does not persist files	heavier for heavy synthesis
writer	"great research, no record" — persists findings.md after skeptic convergence	heavier for exact citation-preserving writing
auditor	"passed because nothing was checked" — evaluates findings.md after the writer	mid — checking, not generating

Model is guidance, not law — chosen per agent by task difficulty, validated at the confirm gate.

Topology — the canonical shape:

explorers (n 2–4, pairwise tensioned)
   │ sequential
synthesizer (1 synthesizer) ◀──zig-zag──▶ reviewers (skeptics; robot_talks when the
   ▲                          │        question needs confrontation, not collection)
   └┄┄┄ feedback (conditional) ┄┘
                                      │ sequential
                                      ▼
                                 writer (1 writer) ──sequential──▶ auditor

The feedback back-edge exists only when there is a reviewer group AND material may be missing — never by default. After skeptic convergence, a separate writer persists findings.md; an optional auditor evaluates that file downstream. Final approval follows the router rather than being inferred from the auditor's presence.

Tension — classify every angle on four axes: methodology (empirical / formal / adversarial / historical / computational), source-corpus, attack-vector (the skeptic gate), temporal-prior (modern-only / historical-lineage / mixed).

• Reject before proposing if: all angles share one core noun phrase; all explorers share one methodology or corpus; all skeptics share one gate.
• Green-light when: for every explorer pair you can write "a_i runs [X], a_j runs [Y] on the [axis] axis; a bias in a_i would be exposed by a_j" — and ≥2 distinct axes appear across the group.

2. Find the evidence

Each explorer retrieves — it does not recall. A claim rests on a source the explorer can point to (a URL, a file path, a citation), never on model memory. Recall is a starting hypothesis; the retrieved source is the proof.

Where each explorer looks is set by its tension axes (see Shape it). source-corpus and temporal-prior assign each explorer a distinct search surface so the group covers the space instead of overlapping — e.g. one sweeps current literature, one the foundational lineage, one the internal repos for prior art.

Two directions, both run before a candidate is judged:

• External — locate with WebSearch, then read the actual source with WebFetch. A snippet is not a read. This is where papers and references come from.
• Internal (ownership) — search our own repos with Grep/Glob (and the inventory skill where installed) for an existing owner. This is what the precedent gate stands on: "already owned here" is retrieved, not assumed.

Each explorer returns the locator per candidate — what it found and where. A sweep that names its surface and tools and comes back empty certifies precedent-clean; an unnamed "found nothing" certifies nothing.

3. Run it

Spawn each group's agents with the Agent tool — ALL agents of a group in ONE message, so they run in parallel. Each agent's initial_prompt is its launch prompt.

Schedule groups by dependency: a group is READY when every group with a sequential/zig-zag edge into it has produced what it must respond to (zig-zag counts only in its from→to direction — the from endpoint opens the exchange). Launch all READY groups concurrently; independent chains run side by side. feedback edges never count as dependencies; a sheet with no connections declares its groups independent. Declared order is a narration tiebreak only.

4. Write the outputs

Results land in the working_folder (a docs path, confirmed at the gate). Which files depends on how many agents ran:

agents	files
2 or more	research.md + findings.md
1	findings.md only

• research.md — the agent returns pasted in verbatim, after a one-section ## Objective preamble. Wrap every return in a uniquely identified stable heading and preserve its original source locators. The strategist writes it by hand; it is transcription, so never dispatch an agent to do it. → How: the domainspec-research-writing skill (.claude/skills/custom/domainspec-research-writing.md)
• findings.md — the synthesis that turns those returns into usable claims, opening Objective → Results → Context. Every load-bearing claim cites the return it rests on. The downstream writer persists it after the synthesizer↔skeptic exchange converges. → How: the domainspec-findings-writing skill (.claude/skills/custom/domainspec-findings-writing.md)
• One-agent dispatches — persist original citations and reproducible details for every executable observation or independent recomputation directly in findings.md; never cite an agent return that was not persisted.

When research-initial-definitions.md registers Research Questions, put a Research-question coverage section in findings.md before the candidate matrix. Include one row for every registered RQ in the confirmed dispatch scope and one explicit row for every registered RQ outside it; never omit an RQ silently. Mark an RQ outside the confirmed dispatch subset deferred and state why that dispatch excluded it, unless an authoritative scope decision retired it; then mark it retired and cite that decision.

RQ id	status	answer	addressable evidence	contrary evidence / material uncertainty	boundary

Apply these rules to each row:

• Use answered, unresolved, deferred, or retired as the status. Use answered only when the cited evidence resolves the entire RQ within its exact confirmed scope. If support is partial, use unresolved, record the supported partial conclusion, and name the exact residual gap. Never record a positive answer without evidence.
• Cite every load-bearing claim. Label support as documentary assertion, executable observation, independent recomputation, or formal proof; do not treat these evidence classes as interchangeable.
• Make findings.md self-contained enough to audit the answer. Cite original sources when available; when relying on collected material, cite its stable return ID in research.md and the original source when available, without duplicating the transcript. Treat research.md as a raw evidence trail, not as accepted evidence by itself.
• For unresolved, cite the material inspected and checks attempted, then name the exact remaining gap. Do not infer absence from failure to find evidence.
• For deferred, retain the reason. For retired, cite the authoritative scope decision.
• State material contrary evidence, residual uncertainty, and the boundary beyond which the answer does not apply; use an explicit none found in <scope> only when the inspected scope is named.

Keep RQ coverage orthogonal to candidate judgment. After the coverage section, put one verdict row per candidate. Ownership is a label, not a verdict: the owner column is always filled (a citation, or precedent-clean), and being owned never puts KILL in the verdict column.

candidate	owner (precedent)	witnessed? (non-vacuity)	sound? (definitional)	verdict	use-mode

• GO — witnessed and sound. use-mode says how: build-from-owned (owned but unused — name the owner + the artifact/job it builds; cite honestly, never claim novel), already-deployed (owned and already wired — provenance only), or novel-attempt (precedent-clean — name the claim, its anchor, and the first obligation a follow-up faces). An owned-but-unused result is a GO, not a negative.
• KILL — only no-witness (non-vacuity) or tautological (definitional collapse); bank it as a typed negative: what it would have contributed + the exact fact that zeroed it. Owned is not a KILL. A clean KILL is a successful run.

5. Close

Close with the one-line answer to the dispatch goal. The dispatch is resolved when the final_approver accepts — and for research, acceptance includes checking that the findings citations hold. Require the auditor and final approver to verify per-RQ coverage, evidence, status discipline, and boundaries whenever Research Questions were registered.

Standing rules

1. Claim ≤ proof — for research, demote, never inflate.
2. Keystone claims carry their collapse-test inline — the one fact that would zero the claim, on the same line.
3. Precedent-first — no novel verdict ships before a precedent skeptic ran. A found owner is not a kill; it relabels the candidate build-from-owned (cite, deploy, never claim novel), and every artifact touching an owned result carries its owner label.
4. Read-only by default — research agents write only into working_folder, never the source tree.

Names

Draw agent_name from telemetry/agents/agent-pool.yaml (ordered role_fit). Prefer the primary role_fit entry and a field fit to the corpus. Never reuse a name within one dispatch (the skeptic/auditor prohibition is the hard case). Never invent a name outside the pool.

See also

• Router — .claude/skills/domainspec-subagents-strategy/SKILL.md: triggers, human gate, anti-bias principle, lifecycle, final_approver, exit_reason vocabulary. Nothing here overrides it.
• Record/sheet mechanics + field definitions — register-dispatch (.claude/skills/register-dispatch/SKILL.md): the two appends, the appender, validation, enums.