| name | yolo-negotiation |
| description | Negotiate a contract with full autonomy -- no checkpoints, no gates. Provide a Word document and instructions. Claude reads, decides, and executes the full pipeline without stopping. Configuration files still inform judgment but never pause execution.
|
| disable-model-invocation | true |
Autonomous Contract Negotiation
Full-autonomy negotiation command. Same negotiation logic as /negotiate but
with all checkpoints removed. Claude reads the document, makes every decision
autonomously, and executes the full pipeline without stopping.
Configuration files (persona, authority, playbook) are still loaded and used to
inform Claude's judgment -- they shape decision quality but never gate
execution.
Step 1: Gather Context
From the user's message, determine:
- Document path -- must be a
.docx file
- Instructions -- what the user wants (brief or detailed)
- Client author name -- for Track Changes attribution. Infer from config
or the user's message. If unclear, ask once.
- Client role -- buyer, seller, landlord, etc. Infer from the instructions
or document context.
Do NOT run a multi-step wizard. Extract what you can from the initial message.
If something critical is missing, ask in one follow-up at most.
Step 2: Load Configuration
Follow the negotiate-contract skill's Step 1 for configuration loading. The
three-level fallback applies:
- Project directory --
PERSONA.md, AUTHORITY.md, PLAYBOOK-*.md,
LOA.md in the current working directory
- Global config --
~/.config/claude-negotiator/
- Shipped defaults -- the plugin's
defaults/ directory
If no custom config exists, the defaults work out of the box. Users can
customise behaviour by placing PERSONA.md, AUTHORITY.md, or
PLAYBOOK-*.md files in their project directory or
~/.config/claude-negotiator/.
Step 3: Ingest and Detect
Call the ingest_document MCP tool with the document path. Then determine the
document type by checking the annotated text for CriticMarkup tracked change
markers:
- No
{++ or {-- markers -- this is a clean document. A document
with only comment markers ({>>) but no tracked change markers is also
treated as clean.
{++ or {-- markers present -- this document has existing tracked
changes.
Step 4: Route to Workflow
Path A -- Clean Document (First-Pass Redlining)
Follow the negotiate-contract skill's First-Pass Redlining Workflow
(Steps A through H) with these modifications:
- Step C (Authority Check): Classify changes against the authority
framework for decision quality, but do NOT pause for amber or red zone
items. Note them in the final report instead. Proceed immediately.
- Step E (Commenting Rules): Same rules apply unchanged.
Path B -- Document with Tracked Changes (Counterparty Response)
Follow the negotiate-contract skill's Counterparty Response Workflow
(Steps 3 through 10) with these modifications:
- Step 3a (Comparison Report): Show a brief inline summary of what the
counterparty did (accepted, pushed back, added new) but do NOT wait for
user confirmation. Display it and immediately proceed to evaluation.
- Step 5a (Authority Check): Classify decisions against the authority
framework for decision quality, but do NOT pause for amber or red zone
items. Note them in the final report instead. Proceed immediately.
- Step 7 (Autonomy Mode): Always fully autonomous. Do not offer supervised
mode. Build the full decision list and execute the pipeline in one go.
Differences from /negotiate
| Gate | /negotiate | /yolo-negotiation |
|---|
| Comparison report | Mandatory wait for confirmation | Shown inline, no wait |
| Authority zones (amber) | Pause and ask for guidance | Note in report, proceed |
| Authority zones (red) | Escalate immediately, do not act | Note in report, proceed |
| Supervised mode | Available for complex negotiations | Never offered |
All other negotiation logic is identical: materiality test, commenting rules,
layering behaviour, edit precision, styler pass, and MCP tool usage.
Quick Reference
MCP Tools
| Tool | Purpose |
|---|
ingest_document | Read clean and annotated text views |
get_state_of_play | Get every change and comment with IDs |
execute_pipeline | Run the full negotiation pipeline (counterparty response) |
redline_document | Apply tracked changes to a clean document (first-pass) |
accept_changes | Accept specific tracked changes (granular) |
counter_propose_changes | Layer counter-proposals (granular) |
add_comments | Add standalone comments (granular) -- use ooxml:NNN anchors after mutations |
reply_to_comments | Reply to comment threads (granular) |
resolve_comments | Mark comment threads resolved (granular) |
Key Rule
Never reject a tracked change. Always layer counter-proposals on top of the
counterparty's markup. The counterparty must see their original change
alongside your response -- full audit trail, full transparency.