skill-reviser

name	skill-reviser
description	Thin wrapper that delegates plan revision to reviser-agent subagent. Invoke for /revise command.
allowed-tools	Agent, Bash, Edit, Read, Write, Glob, Grep

Reviser Skill

Thin wrapper that delegates plan revision to reviser-agent subagent.

IMPORTANT: This skill implements the skill-internal postflight pattern. After the subagent returns, this skill handles all postflight operations (status update, artifact linking, git commit) before returning. This eliminates the "continue" prompt issue between skill return and orchestrator.

Context References

Reference (do not load eagerly):

• Path: .claude/context/formats/return-metadata-file.md - Metadata file schema
• Path: .claude/context/patterns/postflight-control.md - Marker file protocol
• Path: .claude/context/patterns/file-metadata-exchange.md - File I/O helpers
• Path: .claude/context/patterns/jq-escaping-workarounds.md - jq escaping patterns (Issue #1132)
• Path: .claude/context/formats/plan-format.md - Plan file format specification

Note: This skill is a thin wrapper with internal postflight. Context is loaded by the delegated agent.

Trigger Conditions

This skill activates when:

• /revise command is invoked
• Task exists in state.json

---

Execution Flow

Stage 1: Input Validation

Validate required inputs:

• task_number - Must be provided and exist in state.json

# Lookup task
task_data=$(jq -r --argjson num "$task_number" \
  '.active_projects[] | select(.project_number == $num)' \
  specs/state.json)

# Validate exists
if [ -z "$task_data" ]; then
  return error "Task $task_number not found"
fi

# Extract fields
task_type=$(echo "$task_data" | jq -r '.task_type // "general"')
status=$(echo "$task_data" | jq -r '.status')
project_name=$(echo "$task_data" | jq -r '.project_name')
description=$(echo "$task_data" | jq -r '.description // ""')

No status-based ABORT rules. The skill works regardless of task status. Routing is determined by plan file existence, not status.

---

Stage 2: Preflight

No intermediate "revising" status is needed for revision. The task transitions directly to "planned" on success (via postflight). Skip preflight status update.

Rationale: Unlike /plan which sets "planning" as an intermediate status, /revise is lightweight enough that an intermediate status adds no value. The postflight script handles the final status update.

---

Stage 3: Create Postflight Marker

Source skill-base.sh once, then follow @.claude/context/patterns/skill-preflight-flow.md's Stage 3 (marker creation). operation stays "revise" here (not remapped to research/plan/ implement): unlike Stage 2's preflight status update (intentionally skipped above, so there is no update-task-status.sh vocabulary constraint in play for this skill), the marker's operation field is an opaque string with no vocabulary requirement — see skill_create_postflight_marker's signature in skill-base.sh.

source .claude/scripts/skill-base.sh
padded_num=$(printf "%03d" "$task_number")
skill_name="skill-reviser"
operation="revise"
skill_create_postflight_marker "$padded_num" "$project_name" "$session_id" "$skill_name" "$operation"

---

Stage 3a: Calculate Artifact Number

Read next_artifact_number from state.json. Revision uses the current counter directly (not current-1), since each revision is a new planning attempt that advances the sequence.

# Read next_artifact_number from state.json
next_num=$(jq -r --argjson num "$task_number" \
  '.active_projects[] | select(.project_number == $num) | .next_artifact_number // 1' \
  specs/state.json)

# Fallback for legacy tasks: count existing plan artifacts
if [ "$next_num" = "null" ] || [ -z "$next_num" ]; then
  padded_num=$(printf "%03d" "$task_number")
  count=$(ls "specs/${padded_num}_${project_name}/plans/"*[0-9][0-9]*.md 2>/dev/null | wc -l)
  next_num=$((count + 1))
fi

# Revision uses the current counter directly (each revision is a new planning attempt)
artifact_number=$next_num

# Reconciliation: scan all task subdirs for max artifact number on disk
# Handles legacy tasks where next_artifact_number may be behind actual files
padded_num=$(printf "%03d" "$task_number")
max_on_disk=$(find "specs/${padded_num}_${project_name}" -name "[0-9][0-9]_*.md" 2>/dev/null \
  | sed 's|.*/\([0-9][0-9]\)_.*|\1|' | sort -n | tail -1)
max_on_disk=${max_on_disk:-0}
# Strip leading zeros to avoid octal interpretation
max_on_disk=$((10#$max_on_disk))
if [ "$artifact_number" -le "$max_on_disk" ]; then
  artifact_number=$((max_on_disk + 1))
fi

artifact_padded=$(printf "%02d" "$artifact_number")

# Collision check: ensure no existing file uses this prefix in plans/
while ls "specs/${padded_num}_${project_name}/plans/${artifact_padded}_"*.md 2>/dev/null | grep -q .; do
  artifact_number=$((artifact_number + 1))
  artifact_padded=$(printf "%02d" "$artifact_number")
done

Note: Revision increments next_artifact_number in postflight (Stage 8a), so each revision gets a unique, monotonically increasing number.

---

Stage 4: Research Discovery

Discover existing plan and new research reports:

4a. Find existing plan:

padded_num=$(printf "%03d" "$task_number")
plan_dir="specs/${padded_num}_${project_name}/plans"
existing_plan=$(ls -1t "$plan_dir"/*.md 2>/dev/null | head -1)

4b. Find new research reports:

Use BOTH reports_integrated from state.json and file modification time comparison:

reports_dir="specs/${padded_num}_${project_name}/reports"
new_reports=()

if [ -n "$existing_plan" ]; then
  plan_mtime=$(stat -c %Y "$existing_plan")

  # Check reports_integrated from state.json (primary)
  integrated=$(jq -r --argjson num "$task_number" \
    '.active_projects[] | select(.project_number == $num) | .plan_metadata.reports_integrated // [] | .[]' \
    specs/state.json 2>/dev/null)

  for report in "$reports_dir"/*.md; do
    if [ -f "$report" ]; then
      report_basename=$(basename "$report")
      # Check if already integrated
      if echo "$integrated" | grep -qF "$report_basename"; then
        continue
      fi
      # Fallback: check modification time
      report_mtime=$(stat -c %Y "$report")
      if [ "$report_mtime" -gt "$plan_mtime" ]; then
        new_reports+=("$report")
      fi
    fi
  done
else
  # No plan exists -- all reports are "new"
  for report in "$reports_dir"/*.md; do
    if [ -f "$report" ]; then
      new_reports+=("$report")
    fi
  done
fi

---

Stage 4b: Read and Inject Format Specification

Read the plan format file and prepare it for injection into the subagent prompt. This ensures the subagent always has the full format specification in its context.

format_content=$(cat .claude/context/formats/plan-format.md)

The format content will be included as a delimited section in the Stage 5 prompt.

---

Stage 5: Prepare Delegation Context and Invoke Subagent

CRITICAL: You MUST use the Agent tool to spawn the subagent.

Prepare delegation context:

{
  "session_id": "sess_{timestamp}_{random}",
  "delegation_depth": 1,
  "delegation_path": ["orchestrator", "revise", "skill-reviser"],
  "timeout": 1800,
  "task_context": {
    "task_number": N,
    "task_name": "{project_name}",
    "description": "{description}",
    "task_type": "{task_type}"
  },
  "artifact_number": "{artifact_padded from Stage 3a}",
  "existing_plan_path": "{path to existing plan or null}",
  "new_research_paths": ["{path to report1}", "{path to report2}"],
  "revision_reason": "{optional user reason}",
  "roadmap_path": "specs/ROADMAP.md",
  "metadata_file_path": "specs/{NNN}_{SLUG}/.return-meta.json"
}

Required Tool Invocation:

Tool: Agent (NOT Skill, NOT Plan)
Parameters:
  - subagent_type: "reviser-agent"
  - prompt: [Include task_context, delegation_context, existing_plan_path, new_research_paths,
             revision_reason, metadata_file_path,
             AND the format specification from Stage 4b as shown below]
  - description: "Execute plan revision for task {N}"

Format Injection: Include the format specification from Stage 4b in the prompt as a clearly-delimited section:

<artifact-format-specification>
## CRITICAL: Plan Format Requirements

You MUST follow this format specification exactly when writing the plan artifact.
Non-compliance will be caught by postflight validation.

{format_content from Stage 4b}
</artifact-format-specification>

Place this section AFTER the delegation context JSON and BEFORE any other instructions.

DO NOT use Skill(reviser-agent) - this will FAIL.

The subagent will:

• Load revision context files
• Determine revision mode (plan revision or description update)
• Load existing plan and new research reports
• Synthesize revised plan or update description
• Write metadata to specs/{NNN}_{SLUG}/.return-meta.json
• Return a brief text summary (NOT JSON)

---

Stage 5b: Self-Execution Fallback

CRITICAL: If you performed the work above WITHOUT using the Agent tool (i.e., you read files, wrote artifacts, or updated metadata directly instead of spawning a subagent), you MUST write a .return-meta.json file now before proceeding to postflight. Use the schema from return-metadata-file.md with the appropriate status value for this operation.

If you DID use the Agent tool, skip this stage -- the subagent already wrote the metadata.

---

Postflight (ALWAYS EXECUTE)

The following stages MUST execute after work is complete, whether the work was done by a subagent or inline (Stage 5b). Do NOT skip these stages for any reason.

Stage 6: Read Metadata File

Read the metadata file:

metadata_file="specs/${padded_num}_${project_name}/.return-meta.json"

if [ -f "$metadata_file" ] && jq empty "$metadata_file" 2>/dev/null; then
    status=$(jq -r '.status' "$metadata_file")
    artifact_path=$(jq -r '.artifacts[0].path // ""' "$metadata_file")
    artifact_type=$(jq -r '.artifacts[0].type // ""' "$metadata_file")
    artifact_summary=$(jq -r '.artifacts[0].summary // ""' "$metadata_file")
    new_description=$(jq -r '.metadata.new_description // ""' "$metadata_file")
else
    echo "Error: Invalid or missing metadata file"
    status="failed"
fi

---

Stage 6a: Validate Artifact Content

If subagent status is "planned" and artifact_path is non-empty, validate the plan artifact against format requirements. This is non-blocking -- warnings are logged but do not prevent postflight from completing.

if [ "$status" = "planned" ] && [ -n "$artifact_path" ] && [ -f "$artifact_path" ]; then
    echo "Validating plan artifact..."
    if ! bash .claude/scripts/validate-artifact.sh "$artifact_path" plan --fix; then
        echo "WARNING: Plan artifact has format issues (non-blocking). Review output above."
    fi
fi

---

Stage 7: Postflight Status Update

For Plan Revision (status == "planned"):

Update task status to "planned" using the centralized script:

bash .claude/scripts/update-task-status.sh postflight $task_number plan $session_id

If the script exits non-zero, log error but continue (status update is best-effort for revise).

For Description Update (status == "description_updated"):

Update state.json description and TODO.md directly:

Fold --regen-todo in — this write is immediately followed by nothing but the TODO.md regen:

bash .claude/scripts/state-write.sh \