tdk-retro-propose

/tdk-retro-propose - Propose Learning Delta

Error Handling

If prerequisite resolution fails and no local example fixture exists, stop. Do not guess feature paths.

Purpose

Transform collected feedback into up to 10 reviewable learning delta entries. This skill proposes changes only; it does not edit target files.

References

• ../_shared/learning-delta-schema.md
• ../_shared/signal-target-routing.md
• ../_shared/consumer-skill-discovery.md
• ../_shared/script-command-contract.md

Step 1: Resolve Feature Directory

Parse $ARGUMENTS as TASK_ID.

PROJECT_DIR="${CLAUDE_PROJECT_DIR:-${GITHUB_WORKSPACE:-$(git rev-parse --show-toplevel 2>/dev/null)}}"
if [ -z "$PROJECT_DIR" ]; then
  echo "Cannot resolve project root. Run from a git workspace or set CLAUDE_PROJECT_DIR/GITHUB_WORKSPACE."
  exit 1
fi
(cd "$PROJECT_DIR/.specify/scripts/ts" && bun src/commands/util/check-prerequisites.ts {task_id} --paths-only --json)

Store FEATURE_DIR from the JSON output.

Fixture fallback: if path resolution fails and $PROJECT_DIR/.specify/examples/specs/{TASK_ID}/ exists, use that path as FEATURE_DIR. Do not require plan.md; this skill only requires retro-feedback.md.

Step 2: Read Feedback

Read {FEATURE_DIR}/retro-feedback.md. If missing, stop and ask the user to run /tdk-retro-collect {TASK_ID} first.

When reading ## From: User Feedback, use only entries with status: active; ignore entries with status: removed.

Step 3: Discover Consumer Test Skills

Follow ../_shared/consumer-skill-discovery.md:

.claude/skills/*-ut/SKILL.md
.claude/skills/*-test/SKILL.md

If no consumer test skills exist, do not create T4 entries. Record the no-op in rationale when relevant.

Step 4: Check Memory Availability

Check:

.specify/memory/memory-index.md
.specify/memory/memory.yaml

If either is missing, ask:

Memory is not initialized. Run /tdk-memory-init before proposing memory entries?

If the user declines or memory remains missing, skip K1/K2 targets and propose only technical targets T1-T6.

Step 5: Classify Signals

For each feedback signal:

1. Classify root cause with ../_shared/signal-target-routing.md.
2. Choose the narrowest target that prevents recurrence.
3. For memory targets, verify the domain exists in memory-index.md.
4. If domain is unknown, create a blocked entry that says to run /tdk-memory-init.
5. Limit output to 10 entries, highest severity first.

Step 6: Write learning-delta.md

Write {FEATURE_DIR}/learning-delta.md using ../_shared/learning-delta-schema.md.

Required fields per entry:

• status
• severity
• target_type
• target_id
• target_path
• operation
• domain for memory entries
• rationale
• evidence
• content

Completion

Report:

• Entry count by target type
• Blocked entries
• Output path
• Next command: /tdk-retro-apply {TASK_ID}