Ace-Proposal-Respond
Respond to client proposals by converting materials to memory, creating a response strategy, and answering questions in small batches. Uses abd-context-to-memory for RAG. Same iterate-on-strategy pattern as abd-shaping. _
When to Apply
- User wants to respond to an RFP, Q&A, or proposal requirements
- Creating a response plan from proposal documents
- Answering client questions using memory (RAG)
- Iterating on strategy with corrections (DO/DO NOT)
Dependency: abd-context-to-memory
Run index_memory.py --path <proposal_source> before answering questions. Use search_memory.py "<query>" when drafting answers.
Process
- Setup — Convert proposal to memory; create response folder; symlink
- Strategy first — Analyze documents; propose response plan; save to
response/strategy.md - Answer a few questions ONLY — 3–5 per batch; use RAG; get approval
- Accelerators — When answers reference
*See Appendix X (Name)*, define and accumulate in the Accelerator Table (add/update row: slide file, numbers, URL). When done, runbuild_appendix_deck.pyto assemble the appendix deck. - Iterate — Corrections → add DO/DO NOT to strategy; re-run or proceed
Operations
| Operation | When |
|---|---|
create_strategy | "Create strategy," "propose response plan," "analyze and plan" |
answer_questions | "Answer questions," "answer a few," "next batch" |
improve_strategy | "Correct," "fix that," "wrong" |
proceed_slice | "Proceed," "expand," "next slice" |
Scripts
setup_response.py --proposal <folder>— Create response folder and symlinkbuild_appendix_deck.py --table <Accelerator_Table.md> [--output <path>]— Assemble appendix deck from Accelerator Table (requires appendix_config.json or env)
Build
cd skills/abd-proposal-respond
python scripts/build.py