x-internal-map-epic
Generate Implementation Map from Epic and Stories: dependency matrix, phase computation, critical path, ASCII diagrams, Mermaid graphs, strategic observations. Invoked only by x-create-feature (Phase 4).
- License 许可
- MIT
- 最近更新
- 2026-10-27
- Stars
- ★ 0
- Forks
- 1
01 前言 Front Matter
- name
- x-internal-map-epic
- model
- sonnet
- visibility
- internal
- description
- Generate Implementation Map from Epic and Stories: dependency matrix, phase computation, critical path, ASCII diagrams, Mermaid graphs, strategic observations. Invoked only by x-create-feature (Phase 4).
- allowed-tools
- Read, Write, Edit, Bash, Grep, Glob, Skill
- context-budget
- medium
- user-invocable
- false
- requires-capabilities
- []
02 资源文件 1
- 参考文档README.md1.9 KB
03 SKILL.md 正文
🔒 INTERNAL SKILL — Invoked only by other skills via the Skill tool. Not user-invocable.
Output Policy
- Language: Portuguese (pt-BR) for all content. English for technical terms (cache, timeout, handler, endpoint) and code identifiers.
- Tone: Technical, Direct, and Concise.
- Efficiency: Remove all conversational fillers and greetings to save tokens.
Skill: Implementation Map Generator
Purpose
Take the Epic and all Story files and compute the implementation plan: which stories can run in parallel, what the minimum implementation time is, where the bottlenecks are, and how to optimize team allocation. Produce ASCII phase diagrams, Mermaid dependency graphs, and strategic observations for sprint planning.
Triggers
/x-epic-map <epic_file>— generate implementation map from epic and stories- User asks to create an implementation map, generate a dependency graph, or compute implementation phases
- User mentions sequencing stories, finding bottlenecks, computing parallel work streams, or building a roadmap from an epic
Parameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
<EPIC_FILE> |
Path | Yes | — | Path to the Epic file (with story index and dependency declarations) |
--dry-run |
Boolean | No | false | When true, the Implementation Map is written to disk but Steps P4 / P5 (planning-commit / push) become no-ops with a "dry-run, skipping commit" warning (EPIC-0049 / RULE-007). |
Prerequisites
Read the following files before starting:
Template (output structure):
.claude/templates/_TEMPLATE-IMPLEMENTATION-MAP.md— The exact structure to follow
Required inputs:
- The Epic file (with story index and dependency declarations)
- All Story files (with their Blocked By / Blocks tables)
Workflow
Step P1 — Detect Worktree Context (EPIC-0049 / RULE-001)
<!-- TELEMETRY: phase.start -->
Bash command: $CLAUDE_PROJECT_DIR/.claude/hooks/telemetry-phase.sh start x-epic-map Phase-P1-Worktree-Detect
Invoke x-manage-worktrees in detect-context mode so downstream steps know whether the current checkout is already inside an epic worktree. Result is advisory only — x-internal-ensure-epic-branch (Step P2) owns the authoritative branch decision.
Skill(skill: "x-manage-worktrees", args: "detect-context")
Fail-open (RULE-006): any detect-context failure is logged and Step P2 proceeds.
<!-- TELEMETRY: phase.end -->
Bash command: $CLAUDE_PROJECT_DIR/.claude/hooks/telemetry-phase.sh end x-epic-map Phase-P1-Worktree-Detect ok
Step P2 — Ensure epic/<ID> Branch (EPIC-0049 / RULE-001)
<!-- TELEMETRY: phase.start -->
Bash command: $CLAUDE_PROJECT_DIR/.claude/hooks/telemetry-phase.sh start x-epic-map Phase-P2-Epic-Branch-Ensure
Derive the epic ID from the <EPIC_FILE> path (ai/epics/epic-<XXXX>/epic-<XXXX>.md), then ensure the epic/<XXXX> branch exists locally AND on origin. The ensure skill is idempotent — a no-op when the caller is already on that branch.
Skill(skill: "x-internal-ensure-epic-branch", args: "--epic-id <XXXX>")
On non-zero exit, abort with EPIC_BRANCH_ENSURE_FAILED.
<!-- TELEMETRY: phase.end -->
Bash command: $CLAUDE_PROJECT_DIR/.claude/hooks/telemetry-phase.sh end x-epic-map Phase-P2-Epic-Branch-Ensure ok
Step 1 — Build the Dependency Matrix
<!-- TELEMETRY: phase.start -->
Bash command: $CLAUDE_PROJECT_DIR/.claude/hooks/telemetry-phase.sh start x-epic-map Phase-1-DAG-Build
Read every story's Section 1 (Dependencias) and the Epic's story index. Also read each
story's **Chave Jira:** field (if present). Build a complete matrix:
| Story | Titulo | Chave Jira | Blocked By | Blocks | Status |
|---|
The Chave Jira column is placed between Titulo and Blocked By. If a story does not
have a Jira key (field is — or <CHAVE-JIRA>), set the column value to —.
Validation checks:
- Every story in the Epic's index must appear in the matrix
- Dependencies must be symmetric: if A blocks B, then B must list A as blocker
- No circular dependencies (A->B->C->A is invalid)
- Root stories (no blockers) must exist — if every story has a blocker, something is wrong
If inconsistencies are found, fix them and note the corrections.
Add a > **Nota:** block for any implicit dependencies not declared in the stories but
functionally required (e.g., "story-0001-0009 provides configuration data that story-0001-0002 needs even
though story-0001-0002 does not explicitly declare it").
<!-- TELEMETRY: phase.end -->
Bash command: $CLAUDE_PROJECT_DIR/.claude/hooks/telemetry-phase.sh end x-epic-map Phase-1-DAG-Build ok
Step 2 — Compute Phases
<!-- TELEMETRY: phase.start -->
Bash command: $CLAUDE_PROJECT_DIR/.claude/hooks/telemetry-phase.sh start x-epic-map Phase-2-Phase-Computation
Group stories into phases using the dependency DAG:
- Phase 0: All stories with no dependencies (roots)
- Phase 1: Stories whose dependencies are all in Phase 0
- Phase N: Stories whose dependencies are all in Phase 0..N-1
Within each phase, all stories can run in parallel.
Create the ASCII phase diagram using box-drawing characters. Follow the exact style from
the template — +=+, |, +=+ for phase boxes, +-+, |, +-+ for story boxes,
|-->, v for arrows.
Each story box shows: ID + short scope description (max ~20 chars). Each phase box shows: phase number + name + "(paralelo)" if multiple stories.
<!-- TELEMETRY: phase.end -->
Bash command: $CLAUDE_PROJECT_DIR/.claude/hooks/telemetry-phase.sh end x-epic-map Phase-2-Phase-Computation ok
Step 3 — Identify the Critical Path
<!-- TELEMETRY: phase.start -->
Bash command: $CLAUDE_PROJECT_DIR/.claude/hooks/telemetry-phase.sh start x-epic-map Phase-3-Critical-Path
The critical path is the longest chain of dependencies from any root to any leaf. Count phases, not individual stories.
Render as a simple ASCII diagram:
story-0001-0001 -+
+---> story-0001-0002 -> story-0001-0003 --+
story-0001-0009 -+ +---> story-0001-0011
story-0001-0002 -> story-0001-0010 --+
Fase 0 Fase 1 Fase 2 Fase 3
State: N phases in the critical path, M stories in the longest chain. Explain the impact: any delay in a critical path story directly delays the final delivery.
<!-- TELEMETRY: phase.end -->
Bash command: $CLAUDE_PROJECT_DIR/.claude/hooks/telemetry-phase.sh end x-epic-map Phase-3-Critical-Path ok
Step 4 — Generate the Mermaid Dependency Graph
<!-- TELEMETRY: phase.start -->
Bash command: $CLAUDE_PROJECT_DIR/.claude/hooks/telemetry-phase.sh start x-epic-map Phase-4-Mermaid
Create a full graph TD with all stories and their dependency edges.
Naming convention: SXXXX_YYYY["story-XXXX-YYYY<br/>Short Title"]
If Jira keys are available, include them in node labels:
SXXXX_YYYY["story-XXXX-YYYY (PROJ-123)<br/>Short Title"]
Phase coloring (use these exact classDef values for consistency):
classDef fase0 fill:#1a1a2e,stroke:#e94560,color:#fff
classDef fase1 fill:#16213e,stroke:#0f3460,color:#fff
classDef fase2 fill:#533483,stroke:#e94560,color:#fff
classDef fase3 fill:#e94560,stroke:#fff,color:#fff
classDef faseQE fill:#0d7377,stroke:#14ffec,color:#fff
classDef faseTD fill:#2d3436,stroke:#fdcb6e,color:#fff
classDef faseCR fill:#6c5ce7,stroke:#a29bfe,color:#fff
Assign classDef by phase. Group edges by phase transition (comment with %% Fase N -> N+1).
<!-- TELEMETRY: phase.end -->
Bash command: $CLAUDE_PROJECT_DIR/.claude/hooks/telemetry-phase.sh end x-epic-map Phase-4-Mermaid ok
Step 5 — Create the Phase Summary Table
| Fase | Historias | Camada | Paralelismo | Pre-requisito |
|---|---|---|---|---|
| 0 | story-0001-0001, story-0001-0009 | Infra + API | 2 paralelas | — |
Include total count: N historias em M fases.
Add notes about transversal phases (QE, Tech Debt) that can execute independently of business phases.
Step 6 — Detail Each Phase
For each phase, create a subsection with:
Table: Story | Escopo Principal | Artefatos Chave
Entregas da Fase N (bullet list of concrete deliverables — what exists after this phase that did not exist before).
Be specific about artifacts: class names, table names, endpoints, configurations, test infrastructure.
Step 7 — Write Strategic Observations
These are the highest-value part of the map. Analyze:
Gargalo Principal: Which story blocks the most others? Investing extra time in it pays off. This is usually the Layer 1 core story.
Historias Folha (sem dependentes): Stories that do not block anything. They can absorb delays without impacting the critical path. Good candidates for junior developers or parallel streams.
Otimizacao de Tempo: Where is parallelism maximized? Which stories can start immediately? How should teams be allocated across phases?
Dependencias Cruzadas: Stories in later phases that depend on stories from different branches of the dependency tree. Identify convergence points.
Marco de Validacao Arquitetural: Which story should serve as the architectural checkpoint before expanding scope? What does it validate (patterns, pipeline, integration)?
Step 8 — Task-Level Dependency Graph
If stories contain a Section 8 (Sub-tarefas / Tasks) with formal task IDs
(TASK-XXXX-YYYY-NNN), build a task-level dependency graph across all stories.
8A — Extract Cross-Story Task Dependencies
For each story, read the task list (Section 8) and their depends on declarations.
Identify cross-story task dependencies — tasks in one story that depend on tasks from
a different story. Build a table:
| Task | Depends On | Story Source | Story Target | Type |
|---|---|---|---|---|
| TASK-XXXX-YYYY-NNN | TASK-XXXX-ZZZZ-MMM | story-XXXX-YYYY | story-XXXX-ZZZZ | data/interface/schema/config |
Type indicates the nature of the dependency:
data— depends on data model or entity from another storyinterface— depends on a port/interface defined in another storyschema— depends on a DB schema or migration from another storyconfig— depends on configuration or infrastructure from another story
8B — Validate RULE-012 (Cross-Story Consistency)
For every cross-story task dependency, validate that the corresponding story-level dependency exists:
| Validation | Action |
|---|---|
| Cross-story task dep without story-level dep | ERROR: "TASK-X depends on TASK-Y but story-A does not depend on story-B" |
| Cycle detected in task graph | ERROR: "Cycle detected: TASK-A -> TASK-B -> TASK-C -> TASK-A" |
| Story-level dep without cross-story task dep | WARNING: "story-A depends on story-B but no cross-story task dependencies found" |
If an ERROR is detected, abort generation and report the inconsistency. If only WARNINGs exist, proceed but include them in the output.
8C — Compute Merge Order via Topological Sort
Apply topological sort to the full task dependency graph (intra-story + cross-story):
- Build the directed graph of all tasks across all stories
- Detect cycles — if found, abort with cycle description
- Compute topological order (deterministic: break ties by task ID)
- Group tasks into execution phases (phase N = tasks whose dependencies are all in phases 0..N-1)
- Tasks in the same phase with no dependencies between them are parallelizable
Produce a merge order table:
| Order | Task ID | Story | Parallelizable With | Phase |
|---|---|---|---|---|
| 1 | TASK-XXXX-YYYY-NNN | story-XXXX-YYYY | TASK-XXXX-ZZZZ-MMM | 0 |
8D — Generate Mermaid Task Dependency Graph
Create a graph LR with tasks as nodes, colored by story:
- Subgraphs group tasks by story, each with a distinct fill color
- Intra-story edges use solid arrows (
-->) - Cross-story edges use dashed arrows (
-.->) with|cross-story|label - Node labels include task ID and short description
Use these story colors (cycle through for more stories):
style story-1 fill:#e8f4fd
style story-2 fill:#fde8e8
style story-3 fill:#e8fde8
style story-4 fill:#fdf8e8
style story-5 fill:#f0e8fd
style story-6 fill:#e8fdfa
Example:
graph LR
subgraph story-XXXX-0001["Story 0001 (Foundation)"]
style story-XXXX-0001 fill:#e8f4fd
T001["TASK-XXXX-0001-001<br/>Domain Model"]
T002["TASK-XXXX-0001-002<br/>Port Interface"]
T001 --> T002
end
subgraph story-XXXX-0002["Story 0002 (State)"]
style story-XXXX-0002 fill:#fde8e8
T003["TASK-XXXX-0002-001<br/>State Record"]
T004["TASK-XXXX-0002-002<br/>State Enum"]
T003 --> T004
end
T002 -.->|cross-story| T003
If stories do not contain formal task IDs, skip this step entirely and note: "Task-level dependency graph skipped — stories do not contain formal task definitions (TASK-XXXX-YYYY-NNN)."
Step 8.5 — Parallelism Conflict Analysis (EPIC-0041)
<!-- TELEMETRY: phase.start -->
Bash command: $CLAUDE_PROJECT_DIR/.claude/hooks/telemetry-phase.sh start x-epic-map Phase-8-5-Parallelism-Eval
Invoke /x-evaluate-parallelism --scope=epic against the epic under analysis to detect file-level collision risks between stories that Step 2 placed in the same phase. The output feeds the "## 8.5 Restrições de Paralelismo" section of the map.
Fail-open behavior (RULE-005 / RULE-006): If the x-evaluate-parallelism skill is not available in the project (skill file missing under .claude/skills/x-evaluate-parallelism/ or src/main/resources/targets/claude/skills/core/plan/x-evaluate-parallelism/) OR the invocation returns a non-zero exit code, DO NOT abort the map generation. Instead, emit section 8.5 with the following body:
## 8.5 Restrições de Paralelismo
> análise pulada — /x-evaluate-parallelism não disponível (RULE-006 fail-open)
and continue to Step 9. Log a WARNING with the reason (skill missing, exit code N, etc.) for operator diagnostics.
Happy-path output (skill available and returned successfully):
## 8.5 Restrições de Paralelismo
> Análise gerada por /x-evaluate-parallelism em <timestamp omitido para determinismo>.
**Conflitos detectados:** <H> hard, <R> regen, <S> soft
### 8.5.1 Pares Serializados Dentro da Fase
| Fase | A | B | Categoria | Motivo |
| :--- | :--- | :--- | :--- | :--- |
| <Phase N> | <story-A> | <story-B> | <hard|regen|soft> | <file shared or regen collision> |
### 8.5.2 Recomendação de Reagrupamento
<Para cada fase com conflitos, descrever a nova ordem após serialização forçada.>
Determinism (RULE-008): The section MUST be byte-identical across re-runs EXCEPT for the timestamp comment on the header line. The same input epic + story set MUST produce the same collision matrix ordering (sort pairs by Phase ASC → Story-A lexicographic → Story-B lexicographic).
Invocation shape: Invoke via the Skill tool (Rule 13 — INLINE-SKILL pattern):
Skill(skill: "x-evaluate-parallelism", args: "--scope=epic --epic <EPIC_FILE>")
Degenerate case: If no conflicts are detected (0 hard, 0 regen, 0 soft), emit section 8.5 with "Conflitos detectados: 0" and omit subsections 8.5.1 / 8.5.2.
<!-- TELEMETRY: phase.end -->
Bash command: $CLAUDE_PROJECT_DIR/.claude/hooks/telemetry-phase.sh end x-epic-map Phase-8-5-Parallelism-Eval ok
Step 9 — Save and Report
Save as IMPLEMENTATION-MAP.md in the same directory as the Epic and Stories (inside ai/epics/epic-XXXX/).
Report: total stories, phases, critical path length, maximum parallelism, main bottleneck.
If task-level dependencies were computed, also report: total tasks, task phases, cross-story dependencies count.
Step P4 — Commit Implementation Map (EPIC-0049 / RULE-007)
<!-- TELEMETRY: phase.start -->
Bash command: $CLAUDE_PROJECT_DIR/.claude/hooks/telemetry-phase.sh start x-epic-map Phase-P4-Planning-Commit
If --dry-run is set, log "dry-run, skipping commit" and skip this step entirely.
Otherwise, delegate the commit to x-commit-planning so the refreshed ai/epics/epic-XXXX/IMPLEMENTATION-MAP.md is versioned on the canonical epic/<ID> branch without triggering the code pre-commit chain:
Skill(skill: "x-commit-planning",
args: "--scope docs --epic-id <XXXX> --paths ai/epics/epic-<XXXX>/IMPLEMENTATION-MAP.md --subject \"update implementation map\"")
Idempotency: when the map is byte-identical to the previously committed version, x-commit-planning returns commitSha=null and noOp=true (silent no-op). No additional diff check is required here.
On COMMIT_FAILED (exit 4), abort with the same error code.
<!-- TELEMETRY: phase.end -->
Bash command: $CLAUDE_PROJECT_DIR/.claude/hooks/telemetry-phase.sh end x-epic-map Phase-P4-Planning-Commit ok
Step P5 — Push to Origin (optional, EPIC-0049)
<!-- TELEMETRY: phase.start -->
Bash command: $CLAUDE_PROJECT_DIR/.claude/hooks/telemetry-phase.sh start x-epic-map Phase-P5-Push
If --dry-run is set, log "dry-run, skipping push" and skip.
Otherwise, push the canonical epic branch so the updated map is observable on origin:
Skill(skill: "x-push-branch", args: "--branch epic/<XXXX>")
On push failure, log a WARNING and continue — the local commit is preserved; the operator re-runs manually. Do NOT abort.
<!-- TELEMETRY: phase.end -->
Bash command: $CLAUDE_PROJECT_DIR/.claude/hooks/telemetry-phase.sh end x-epic-map Phase-P5-Push ok
Error Handling
| Scenario | Action |
|---|---|
| Template file missing | Abort with message: "Template _TEMPLATE-IMPLEMENTATION-MAP.md not found" |
| Epic file missing or unparseable | Abort with message: "Epic file not found or invalid format" |
| Story files missing | Abort with message: "No story files found in epic directory" |
| Circular dependency detected | Warn with cycle details; break the cycle and proceed |
| Asymmetric dependency (A blocks B but B does not list A) | Fix automatically and note the correction |
| No root stories (all have blockers) | Warn: "All stories have blockers — review for missing root stories" |
| Cross-story task dep without story-level dep (RULE-012) | Abort with error listing the inconsistent dependencies |
| Cycle in task dependency graph | Abort with cycle path: "Cycle detected: TASK-A -> TASK-B -> TASK-A" |
| Stories without formal task IDs | Skip Section 8 with note; proceed with story-level map only |
/x-evaluate-parallelism skill missing or exit code ≥ 1 (Step 8.5) |
Fail-open: emit section 8.5 with "análise pulada — /x-evaluate-parallelism não disponível" (RULE-005 / RULE-006); log WARNING; continue to Step 9 |
x-internal-ensure-epic-branch fails (Step P2) |
Abort with EPIC_BRANCH_ENSURE_FAILED; canonical branch is required for versioning |
x-commit-planning exit 4 (Step P4) |
Abort with COMMIT_FAILED; file written but not versioned |
x-commit-planning exit 0 + noOp=true (Step P4) |
Silent no-op — re-execution idempotency confirmed; continue to Step P5 |
x-push-branch fails (Step P5) |
WARN only; local commit preserved; operator re-runs manually |
--dry-run set |
Steps P4 and P5 become no-ops with log line "dry-run, skipping commit" / "dry-run, skipping push" |
Common Mistakes
- Phase computation error: A story can only enter a phase when ALL its dependencies (not just some) are in earlier phases
- Missing convergence analysis: When story-0001-0011 depends on story-0001-0003 AND story-0001-0010 from different branches, this creates a convergence point that deserves a callout
- Generic observations: "story-0001-0002 is important" says nothing. "story-0001-0002 blocks 6 stories and establishes the decision engine pattern that all Phase 2 handlers reuse — investing extra design time here prevents refactoring 6 handlers later" is useful
- Inconsistent status: If a story is marked Done in the matrix but Pending in the phase diagram, that is a bug
- Missing leaf analysis: Leaf stories (no dependents) are strategically important because they can absorb schedule variance. Always identify them
Integration Notes
| Skill | Relationship | Context |
|---|---|---|
| x-epic-create | reads | Reads the Epic file with story index and dependencies |
| x-story-create | reads | Reads generated story files for dependency details |
| x-epic-decompose | called-by | Orchestrator invokes x-epic-map in Phase D |
| x-implement-epic | reads | Epic implementation reads the map for execution order |
| x-manage-worktrees | calls (Step P1) | Detect-context (EPIC-0049 / RULE-001) |
| x-internal-ensure-epic-branch | calls (Step P2) | Ensure epic/<ID> exists locally + origin (EPIC-0049 / RULE-001) |
| x-commit-planning | calls (Step P4) | Batch-commit implementation map without code pre-commit chain (EPIC-0049 / RULE-007) |
| x-push-branch | calls (Step P5) | Push canonical epic branch to origin (optional) |
Knowledge Pack References
| Knowledge Pack | Usage |
|---|---|
| story-planning | Phase computation, dependency DAG, critical path analysis |
| parallelism-heuristics | Step 8.5 collision categories (hard/regen/soft) consumed from /x-evaluate-parallelism output |
Planning Status Propagation (Rule 22 / EPIC-0046)
V2-gated: only runs when the epic declares
planningSchemaVersion: "2.0"in itsexecution-state.json. For v1 epics: skip silently (Rule 19).
x-epic-map reads story files (it does NOT write to them) and emits/refreshes ai/epics/epic-XXXX/IMPLEMENTATION-MAP.md with two lifecycle columns:
- Planejamento — mirrors the current
**Status:**of each story (values:Pendente,Planejada,Em Andamento,Concluída,Falha,Bloqueada). - Status — the execution lifecycle status (same six values; driven by
x-implement-story/x-implement-epicdownstream).
Both columns are populated by reading the story files in situ via the CLI. x-epic-map never transitions any story — it is a read-only skill with respect to lifecycle status. The {{PLANNING_STATUS}} token in the implementation-map template is resolved at render-time by reading each story.
Steps (end of map generation, BEFORE the final commit):
-
Detect v2 via
execution-state.json. If v1: leave legacy columns untouched (skip this block). -
For each story row being rendered into the map:
STATUS=$(java -cp $CLAUDE_PROJECT_DIR/java/target/classes \ dev.iadev.adapter.inbound.cli.StatusFieldParserCli \ read ai/epics/epic-XXXX/story-XXXX-YYYY.md)Use
$STATUSto fill thePlanejamentocolumn for that row. Exit code 20 for any story → abort map generation (source-of-truth invariant RULE-046-01 must hold). -
The
Status(execution) column is filled fromexecution-state.json(not from the story file) — the story's**Status:**line is authoritative for planning only in v2. -
Stage and commit the updated map:
git add ai/epics/epic-XXXX/IMPLEMENTATION-MAP.mdSkill(skill: "x-commit-changes", args: "docs(epic-XXXX): refresh implementation map with lifecycle columns")
Fail-loud: CLI exit 20 on any story aborts map generation (RULE-046-08). x-epic-map never calls write on a story — the skill only reads.
04 同仓库更多 Skills
x-parallel-eval
Detects and classifies file-collision risks between work units (tasks, stories, or epic phases) before parallel execution. Reads ## File Footprint blocks produced by x-task-plan / x-story-plan, applies the parallelism-heuristics knowledge pack (hard / regen / soft categories + hotspot overrides), and emits a collision matrix + serialization recommendation in Markdown (default) or JSON.
parallelism-heuristics
Canonical catalog of file-collision heuristics for parallel task, story, and epic execution. Defines the File Footprint block format, the three Conflict Categories (hard / regen / soft), a hotspot list of high-contention paths, and the degrade-with-warning policy. Consumed by x-task-plan, x-story-plan, x-epic-map, x-parallel-eval, and x-dev-*-implement.
x-watch-pr-ci
Polls a PR's CI checks and Copilot review status, blocking until checks complete or timeout. Returns one of 8 stable exit codes (SUCCESS=0, CI_PENDING_PROCEED=10, CI_FAILED=20, TIMEOUT=30, PR_ALREADY_MERGED=40, NO_CI_CONFIGURED=50, PR_CLOSED=60, PR_NOT_FOUND=70). Writes a versioned state-file for session resume.
x-validate-docs
Documentation freshness gate: validates 6 dimensions (readme, api-specs, grpc-proto, adr, skill-docs, system-architecture) against code changes in a PR. Stack-aware — only validates targets declared/auto-detected from ProjectConfig.documentation.targets. Returns exit non-zero on staleness; produces structured report.
x-update-system-architecture
Incrementally updates docs/architecture/system.md after an epic completes. Appends new entries to the Decision Log (§11) using x-internal-write-report --append (dedup by ## ID: marker) and surgically inserts component/integration changes into sections 1-10 via Edit. Idempotent: re-running with the same epic produces a byte-identical system.md.
x-update-architecture
Incrementally updates the service architecture document with changes from architecture plans. Adds new components, integrations, flows, and ADR references without rewriting existing content. Use after implementation to keep architecture documentation current.