maestro-odyssey

<required_reading> @/.maestro/workflows/run-mode.md @/.maestro/workflows/codex-run-mode.md </required_reading>

If any required file above was not expanded into context by the host, or its content is no longer in context, Read it explicitly before executing any step.

<deferred_reading>

• odyssey-base.md — read after mode resolved for shared back-half (INTAKE gate → GENERALIZE → DISCOVER → RECORD → END)
• odyssey-debug.md — read when mode=debug
• odyssey-improve.md — read when mode=improve
• odyssey-planex.md — read when mode=planex
• odyssey-review.md — read when mode=review
• odyssey-security.md — read when mode=security
• odyssey-defensive.md — read when mode=defensive
• odyssey-ui.md — read when mode=ui </deferred_reading>

<purpose> Long-running, evidence-driven iterative cycle. A single entry dispatches to one of seven modes; all share the same skeleton — discovery → domain audit → fix → verify → generalize → discover siblings → persist knowledge — and iterate exhaustively until the mode's exit condition is met or escalation is required. </purpose>

<mode_dispatch>

Mode selection precedence: explicit --mode <name> > intent keyword auto-detection > request_user_input (Normal) / error E000 (-y).

Auto-detection from <intent> keywords (first match wins, ordered):

Keyword matching: case-insensitive substring match against the intent text. Multi-word keywords require all words present (not necessarily adjacent). First matching row wins (ordered by specificity).

Keywords in intent	Detected mode
bug, crash, error, broken, fails, regression, race, leak, "why does"	debug
requirement, implement, build, add feature, I need to implement, I need to build, I need to add, deliver feature, user story	planex
ui, visual, layout, style, component, page, responsive, a11y, accessibility, UI design, visual design, design system, design tokens	ui
security audit, OWASP, vulnerability, CVE, secrets scan, STRIDE, threat model, supply chain, dependency audit, dependencies, supply chain audit	security
defensive programming, defensive code, exception swallowing, silent failure, fallback risk, default value risk, error suppression, 防御性编程, 防御性代码, 异常吞噬, 兜底风险	defensive
improve, optimize, performance, refactor quality, reliability, observability	improve
review, audit, code check, check the code, inspect the code, inspect changes, "look over", zero-residual	review

Ambiguous / no match → Normal: request_user_input (7-way mode pick) | -y: E000.

Mode registry:

Mode	Purpose	Discovery phases	Audit phase	Fix→verify pair	Unique states
debug	Symptom → root cause → fix → confirm	ARCHAEOLOGY, EXPLORE	DIAGNOSE (hypothesis test)	FIX → CONFIRM	ESCALATE_DIAGNOSIS
improve	6-dimension quality audit → diagnose → fix	SURVEY	AUDIT (6 dims) + DIAGNOSE	FIX → VERIFY	ESCALATE_DIAGNOSIS
planex	Requirement → plan → execute → verify loop	(none)	PLAN + EXECUTE	(EXECUTE) → VERIFY → FIX loop	—
review	Multi-dimension deep review → zero-residual fix	ARCHAEOLOGY, EXPLORE	REVIEW (4+ dims)	FIX → CONFIRM	—
security	Read-only tiered security audit → severity matrix	RECON	SCAN (OWASP + deps + secrets + CI/CD + STRIDE + git)	(none — read-only)	—
defensive	Business-anchor → backward-slice → 8-pattern scan → forward-propagate → risk score	ANCHOR, SLICE	SCAN (8 defensive patterns) + PROPAGATE	(none — read-only)	—
ui	Visual survey → 6-dim audit → diverge → fix	SURVEY	AUDIT (6 dims) + DIVERGE	FIX → VERIFY	—

CONFIRM and VERIFY are synonymous — both refer to the post-fix validation phase. Mode workflow files use mode-specific naming; semantics are identical.

The back half is identical across all modes: GENERALIZE → DISCOVER → RECORD → END (see odyssey-base.md §Shared Back-Half).

On mode resolved: read the deferred workflow file for that mode + odyssey-base.md, then execute.

</mode_dispatch>

<context> $ARGUMENTS

Universal flags: --mode <name> mode selector | --skip-fix audit/diagnose only, skip fix+verify | --skip-generalize skip GENERALIZE+DISCOVER | -y skip all confirmation interactions (including delegate/agent confirmations in execution phases), use default choices; decisions skipped this way are recorded as deferred; never bypasses mode ambiguity (E000), INTAKE gate blockers, escalation | -c resume the most recent unfinished Session of the SAME mode via exact Session resolution: locate it with maestro session list --json + maestro session status --session {session_id} --json (both read-only), re-attach context with maestro session resume-view and the brief-result/3.0 Resume Packet via run brief (exact invocation per run-mode.md), then continue the chain with fenced maestro run next / run check / run complete --advance. If --mode conflicts with the resumed Session's mode → E003 (mode mismatch); no history → ignore -c, create new Session | --heartbeat /loop periodic progress

Mode-scoped flags:

Flag	Modes	Description	Default
--template <name>	debug, planex	Predefined strategy/criteria template	—
--dimensions <list>	improve, review, ui	Audit dimension subset	all
--fix-threshold <sev>	improve, review, ui	Severity cutoff (critical|high|medium|low|all)	all
--tier quick|standard|deep	security, defensive	Audit depth tier	standard
--sink-depth <list>	defensive	Explicit sink layer focus (1|2|3|all)	all
--max-iterations N	planex	Max verify-fix cycles before escalation	3
--method agent|cli|auto	planex	Task execution method	auto
--executor <tool>	planex	Explicit CLI executor	first enabled
--skip-verify	planex	Skip post-execution validation gate	false

--skip-fix applicability: security and defensive modes ignore (read-only, no fix phase); planex skips FIX loop but retains EXECUTE+VERIFY; debug/review/improve/ui skip FIX+VERIFY/CONFIRM. --skip-fix + --skip-verify on planex = PLAN only (no execution).

Mode-scoped flags passed to inapplicable mode: emit W008 warning and ignore the flag.

Session creation: follow run-mode.md exactly. Negotiate capabilities, then use the receipt-chained self-start path: open an empty Session with --actor carrying the authorized identity, insert odyssey-<mode> plus every mode/domain argument through fenced repeatable --arg, and dispatch it with fenced run next using a third request ID. Every mutation consumes the exact revision returned by the preceding receipt. Consume the resulting birth packet's resolved task and structured executable continuation; never pass task prose via --input. A direct machine-protocol run create is allowed only with positional domain arguments, explicit --session/--run/--step, --actor identity, exact CAS revision, and --input values restricted to sealed same-Session Artifact IDs.

Session: {run_dir}/outputs/ Output: session.json | evidence.ndjson | understanding.md | explore.json (debug/review only) | anchors.json (defensive only)

Output boundary: ALL session artifacts MUST target the run outputs directory ({run_dir}/outputs/) only. .workflow/state.json and all sessions/<sid>/ protocol files are runtime-owned — a workflow never writes them. Source code modifications during fix/execute phases are in-scope but MUST be committed per action. NEVER write session artifacts outside {run_dir}/outputs/.

session.json — shared core + mode fields:

{ "mode": "debug|improve|planex|review|security|defensive|ui",
  "target": "", "dimensions": [],
  "patterns": [], "confirmation": null, "generalization_stats": null,
  "cross_phase_loops": 0 }

Each mode extends the core — see the mode's workflow file for session fields.

Commit convention: "odyssey-{mode}({slug}): {STATE} — {summary}" (mode = active mode short name; review mode uses odyssey-review).

</context> <invariants> All base invariants apply (evidence append-only, session-as-state, phase goal tracking, auto-commit per action). Additionally:

1. Evidence append-only — never delete or overwrite evidence.ndjson entries.
2. Phase goal tracking — mark each goal done/failed before transition; no silent skips.
3. Generalize is mandatory — GENERALIZE and DISCOVER execute unless skip_generalize == true. Prior-phase convergence, "no findings / all verified / zero remaining," or context pressure are NOT valid skip reasons. The phase itself determines whether patterns exist.
4. Zero-residual (improve/review/ui) — every finding MUST have a concrete action (fix / issue / decision). "Report and shelve" and blanket "pre-existing" skips are forbidden.
5. Read-only (security, defensive) — NEVER modify source code, configuration, or dependencies. Security and defensive audits produce reports only; fixes route to --mode improve.
6. Acceptance criteria are sacred (planex) — no "close enough", no manual override without explicit escalation.
7. Browser is truth (ui) — verify in real rendering, not just code review. Diverge before converge.
8. Goal 跟宿主原生 /goal 走 — session.json 是权威进度。不要为了刷新外观去重建 Goal；用户 clear 之后外观必须保持消失。 </invariants>

<task_tracking>

时机与操作（plan 是 session 权威状态的 UI 镜像，不替代 session 状态）：

时机	操作	示例
Session 创建后	update_plan 初始化步骤清单	update_plan({ plan: [{ step: "Step {index}: {step.skill}", status: "pending" }, ...] })
Step 派发时	update_plan 标记当前 step	update_plan({ plan: [..., 当前 step status: "in_progress"] })
Step 完成时	update_plan 标记完成	update_plan({ plan: [..., 该 step status: "completed"] })
Step 失败时	update_plan + explanation 说明	update_plan({ explanation: "Step {index} failed: {reason}", plan: [...] })

</task_tracking>

<self_iteration> Self-iteration (logic in odyssey-base.md) applies to each mode's discovery + audit + GENERALIZE stages:

Mode	Self-iterating stages
debug	S_ARCHAEOLOGY, S_EXPLORE, S_DIAGNOSE, S_GENERALIZE
improve	S_SURVEY, S_AUDIT, S_DIAGNOSE, S_GENERALIZE
planex	S_PLAN, S_VERIFY, S_GENERALIZE
review	S_ARCHAEOLOGY, S_EXPLORE, S_REVIEW, S_FIX, S_GENERALIZE
security	S_RECON, S_SCAN, S_GENERALIZE
defensive	S_ANCHOR, S_SLICE, S_SCAN, S_PROPAGATE, S_GENERALIZE
ui	S_SURVEY, S_AUDIT, S_DIVERGE, S_GENERALIZE
</self_iteration>

<execution> Follow base execution discipline completely. On entry: resolve mode (§mode_dispatch), then read the deferred workflow file for that mode + odyssey-base.md, and run that mode's state machine. All modes converge on the Shared Back-Half in odyssey-base.md.

Shared Phase Gates (MANDATORY, BLOCKING)

• INTAKE gate: mode resolved, target/requirement resolved, SESSION_DIR created, session.json initialized (with baseline_metrics for improve; acceptance_criteria for planex), phase_goals[] derived from flags, understanding.md §1 written. BLOCKED if no target (E001) / no requirement (planex E001) / target path not found (E002) / mode unresolved (E000).
• GENERALIZE gate: ALL 3 layers (syntax/semantic/structural) attempted with evidence logged; generalization_stats written with by_layer entries for all 3 layers; generalize goal marked. Any layer not attempted = thoroughness-floor violation (BLOCKED).
• DISCOVER gate: all hits triaged with per-item classification and reason; remaining_actionable == 0 OR loops >= max_loops with per-item reasons logged; discover goal marked. Unclassified hits = BLOCKED.

Mode-specific phase gates (Discovery, Audit, FIX, VERIFY/CONFIRM) are defined in each mode's workflow file.

</execution>

<error_codes>

Code	Severity	Condition	Recovery
E000	error	Mode unresolved (-y, ambiguous intent, no --mode)	Provide --mode
E001	error	No target / no requirement (planex) / no issue (debug)	Provide target or -c
E002	error	Target path not found	Check path
E003	error	-c mode mismatch (resumed session is different mode)	Use correct --mode or omit -c
E004	error	Mode workflow file not found (~/.maestro/workflows/odyssey-{mode}.md)	Verify workflow installation or select another mode
W001	warning	No relevant git history / no dependency manifest / no design system	Proceed with defaults
W002	warning	Some dimension agents failed / 3 retries exhausted	Partial coverage / INCONCLUSIVE
W003	warning	Archaeology agent or delegate failure (debug/review)	Proceed with available results, log failed agent
W004	warning	Generalization 0 hits after full 3-layer scan	Advance to S_RECORD (requires all 3 layers attempted with evidence)
W005	warning	Pending decisions	Filter evidence phase=decision
W006	warning	No CLI tools (debug/review explore)	Skip explore
W007	warning	planex CLI review regression concern	Review before next iteration
W008	warning	Mode-scoped flag ignored (not applicable to resolved mode)	Remove flag or use correct mode
</error_codes>

<success_criteria>

• Mode resolved (explicit or auto-detected); session + output files created; prior knowledge searched
• Discovery phase(s) for the mode completed with evidence (archaeology/explore/survey)
• Domain audit completed with structured findings + severity matrix (or acceptance criteria + plan for planex)
• understanding.md sections written progressively per mode
• Fix + verify/confirm (unless --skip-fix); zero-residual for improve/review/ui; all criteria pass for planex
• Read-only invariant maintained for security and defensive modes — zero source modifications
• Multi-layer generalization + discovery triage (unless --skip-generalize); every unfixed finding individually justified
• phase_goals derived, tracked, and hardened-audited; goal_mode injected via prepare goal:true; -y no blocking prompts
• Session resumable via -c; mode-specific completion summary emitted </success_criteria>

<next_step_routing>

Condition	Next
Single-file mechanical fix discovered	/maestro-companion "<fix>"
Discovery issues created	/maestro-issue list --source {mode}-odyssey
Deeper debug needed (from any mode)	/maestro-odyssey <finding> --mode debug
Security findings need remediation	/maestro-odyssey <finding> --mode improve
Defensive findings need remediation	/maestro-odyssey <finding> --mode improve
Defensive root-cause needed	/maestro-odyssey <finding> --mode debug
Formal review of changes	/maestro-odyssey <changed-files> --mode review
UI-related findings	/maestro-odyssey <component> --mode ui
Document pattern	/maestro-learn decompose <module>
Second opinion	/maestro-learn consult <understanding.md>
Related question	/maestro-learn investigate "<question>"
Design/perf/arch pattern to persist	/maestro-spec add ui|coding|arch "..."
Pending decisions	Filter evidence phase=decision status=pending
</next_step_routing>