Module prompt 

Prompt construction.

These strings are the actual product. The graph only moves bytes around; how well a run goes is decided by what the judges are asked to look at and what they are forbidden to speculate about.

Two rules run through all of them:

• No authorship. Nothing an agent receives names a model or a vendor, and every prompt that could invite a guess explicitly forbids guessing.
• Checkable claims. Judges and reviewers are told to verify assertions against the repository, and to name a trigger for every defect. That is what makes an unread patch defensible.

Structs

CandidateView

One candidate as presented to a judge.

ChatTaskNotice

What chat_task_notice reports; every field is plain data magi read.

ConductAnswer

One answered question, for ConductTask::answers and [ConductOutcome::answers].

ConductFinding

One finding, as shown to the conductor across every review round — not only the last one. See ConductOutcome::rounds for why every round matters here.

ConductFinished

A Failed/Held task together with how its last run ended.

ConductOutcome

How a task’s last run ended, for a Failed/Held task the conductor has not yet been shown — the “終わったタスク” the whole feature exists for.

ConductRejection

One finding the fixer declined, and why — see ConductRound::rejected.

ConductRound

One review round’s findings and how the fixer treated each one, for ConductOutcome::rounds.

ConductTask

A task shown to crate::conduct: either runnable (a dependency-blocking target), or Running past the stall threshold with no live daemon claiming it. priority is shown so the conductor can see the order the loop already runs in — never so it can change it: nothing in crate::conduct::Decision carries a priority back.

DeputyPrompt

What a conductor question’s deputy is told.

NoticeDisplay

Display-only snapshot of an automatic event, separate from agent instructions.

RebaseConflict

What the fixer is told when a rebase of the winning branch stopped on a conflict.

ReviewCtx

Everything a reviewer needs to know about the patch under review.

ReviewPatch

The patch text a stateless reconsideration call restates. See ReviewReconsiderCtx::patch.

ReviewReconsiderCtx

Everything a reviewer needs to reconsider its vote after a split round.

ReviewSeatReport

One reviewer seat’s report, as shown to the rest of the panel during reconsideration. Seats stay numbered, never named — the same convention review itself uses for panel size, not a disclosure of identity.

Turn

A judge’s contribution to the deliberation transcript.

Enums

Lens

One of the fixed angles a reviewer seat is assigned.

OwnerWord

What the owner said, handed to the agent that asked.

Constants

CHANGE_SUMMARY_HEADING

Heading of the whole-change summary prompt (the end-to-end mock greps it).

CHAT_CONSULT_DEFUSED

Opening sentence that only a defused (current) consult carries; the no-break space is what consult::owner_words looks for.

CHAT_CONSULT_END

The last words of every generated consult text, used to find where it ends.

CHAT_CONSULT_HEADING

Heading of the turn that hands a question to the chat it came from.

CHAT_TASK_NOTICE_HEADING

Heading of the automatic turn that reports a chat-filed task’s final state to the conversation. The briefing names it, and consult leaves it out of the owner’s words.

CHAT_WATCH_NOTICE_HEADING

Heading shared by watch draft detection and owner-message filtering.

DEPUTY_HEADING

Heading of the deputy prompt, which the end-to-end mock agent greps for.

DUPES_JUDGE_HEADING

Heading of the duplicate-work judge’s prompt; the test mock agent dispatches on it.

DUPES_JUDGE_MAX_CHARS

Longest instruction the duplicate-work judge is given, in characters. A longer one keeps its head and tail, with the cut marked.

GITHUB_ENGLISH_HEADING

Heading of the fixed rule below; tests and callers key on it.

GITHUB_LANGUAGE_HEADING

Heading of the rule used when [graph] github_language is not English.

MAX_PATCH_BYTES

Patches above this size are truncated in the prompt; the judge is pointed at the branch instead. Agent context windows are large but not free, and a 10 MB vendored-dependency diff is not read by anyone anyway.

QUESTION_RESUMED_HEADING

Heading of question_resumed; the mock agent in tests/common keys on it.

WORK_IN_PLACE_HEADING

Heading of the section that tells a seat not to build a second checkout.

WORK_IN_PLACE_RULE

The one sentence group every seat that works in a magi-made worktree is given. Tool-agnostic on purpose: the repository’s own instructions may name any worktree / workspace helper, and magi has already made the worktree with plain git. It must not name a vendor (the judge prompt is checked for that) nor any heading the test mock dispatches on.

Functions

advisor

Prompt for one advisor seat in the design-deliberation stage (crate::graph::Runner::advise), run before any implementer touches the repository.

build_cache_note

What a seat is told about the shared build cache — one of two notes, chosen by whether the seat may write at all.

change_summary

Ask a read-only agent to describe the whole change between base and head for a reviewer who has seen neither the task nor the conversation. The commit list and diff stat are orientation only; the agent must read the diff itself and the code around it.

change_summary_in

change_summary with [graph] github_language.

chat_consult

The operator turn that puts an open question in front of the chat agent.

chat_consult_display

User-facing consultation snapshot, without the instructions sent to the agent.

chat_task_notice

The automatic message that tells a chat how a task it filed ended.

chat_task_report

Build agent input and display snapshot from the same sanitized facts.

chat_watch_notice

Fixed English report; command output is quoted data, scrubbed and bounded.

chat_watch_report

Build agent input and display snapshot from the same sanitized facts.

child_run_ownership

Fixed guidance for child seats, including paths that bypass the wave launcher. This is prompt guidance, not command enforcement; keep control-plane use valid.

conduct

Prompt for crate::conduct’s single seat.

defuse

Make the consult markers unrecognisable in text that is only quoted.

deliberate

Prompt for one deliberation turn.

deputy

See DeputyPrompt.

dupes_judge

The brief for the one-shot duplicate-work judge. claims are the rendered matches (which task / run / pull request owns what) with the owner’s own work in one line (empty when unknown); instruction is the text of the work being filed. The text is data: the judge is told not to obey anything inside it.

final_vote

Prompt for the private final vote.

fix

Prompt for the fixer, given a round’s findings.

fix_in

fix with [graph] github_language.

gate_fix

Prompt for the fixer when the final verify.gate failed on a tree the reviewers had already cleared.

gate_fix_in

gate_fix with [graph] github_language.

github_english

The rule that everything landing on GitHub is English, whatever [graph] language says and whatever language the task was written in.

github_english_finding_titles

github_english for a reviewer: the only thing of theirs that reaches GitHub is a finding’s title, which the pull request body lists.

github_finding_titles_rule

github_rule for a reviewer’s finding titles.

github_rule

github_english for a configured [graph] github_language. English is byte-for-byte the fixed rule above; any other language names it for pull request / issue titles, bodies and comments, keeps commit messages English and leaves the quality and confidentiality rules untouched.

graft

What the fixer is told in the graft round: port the named strengths of the losing candidates onto the winner, and nothing else.

implement

Prompt for an implementer.

implement_in

implement with [graph] github_language.

is_completion_notice

Recognize automatic completion reports in a conversation message.

judge

Prompt for a blind judge.

legacy_notice_display

Strict compatibility reader, called only after provenance has been verified. Unknown completion templates retain their original text; old consultations use a generic placeholder because no display snapshot was saved.

nudge

Follow-up when a reply could not be parsed.

operator_fix

Prompt for a targeted, operator-triggered fix: specific, already-recorded findings routed to a fixer outside the normal review round sequence.

operator_fix_in

operator_fix with [graph] github_language.

question_resumed

The prompt that resumes a seat whose magi ask is gone, so the owner’s word still reaches the agent that already holds the context.

rebase_conflict

See RebaseConflict.

recovery_fix

The fixer’s prompt for the one repair a verification-only failure earns: fix with the failed verification placed ahead of the task.

recovery_review

The section a reviewer of the first round after a takeover is shown.

resume_after_drop

Follow-up when the CLI hung up before delivering an answer.

resume_incomplete

Follow-up when the CLI’s own turn ended cleanly — a usable, non-empty, exit-0 reply — but held none of the structured report this step reads back.

review

Prompt for a reviewer of the winning patch.

review_reconsider

Prompt for the one round of reconsideration a split review vote earns.

synthesize_brief

Prompt for the synthesis seat that blends the advisors’ proposals into a design brief carried in the implementer’s prompt (crate::prompt::implement’s brief argument).

with_overlay

Append the project’s overlay for a node, under a heading of its own.