Directory
‹ Back to directory

D'Artagnan

Bots

1

An outer loop that sits above coding agents. Name a repo and a goal; it gathers context, writes a testable prompt, launches one cloud agent, reviews the PR, and merges only with proof.

2

An evidence and stamp desk. It checks claims against primary sources and returns aligned, partial, contested or insufficient, and it stamps pull requests or content PASS, REVISE or FAIL with a receipt tied to the exact version it checked.

3

Home cooking and cocktail coach with food-science depth — Western, Eastern, German, French, Mediterranean, Italian kitchens plus classic drinks. Teaches the why, and can build an Obsidian knowledge graph (including Science/) when you want one.

4

Sourdough starter and bake coach. Helps you make, feed, track, and bake with a starter — feed ratios, rise windows, kitchen climate, fridge vs counter, and troubleshooting sluggish jars or dense loaves. Can rebuild a local Obsidian knowledge graph on request.

5

A negotiation and multi-party strategy desk. Bring a high-stakes deal, hire, investment, or partnership. It maps every side's incentives and hands you your best move, then checks your own plan for blind spots.

6

A coding and strategy partner that tells you what the experts would do, then bets on the smart move they wouldn't, with odds, a stake cap and a kill test up front. It runs small testable experiments and keeps only what works.

7

Stops a working bot from grading its own homework. You write a frozen charter; a Worker does the job, a separate Governor audits each run against the charter, and a Curator keeps memory typed and dated. Nothing gets promoted without your yes.

8

For owners running several Grok Bots whose rules overlap and drift. Crewfold digests them into one profile, fleet map, routine list and diff, then stops for your approval before anything changes.

9

A pre-launch gate for apps, sites, and paid plans. Before anything goes live, it runs a checklist covering security, money, legal, and the core user path, and returns GO, GO-WITH-FIXES, or NO-GO with evidence for each item.

10

Reads the launches, tools, and saved links you'd otherwise skim, scores each against your goals and a written bar, and sorts it Keep, Park, or Kill. Quiet unless something clears the bar. Parked items get a revisit date; killed ones stay dead.

11

Sweeps your bot crew for dropped handoffs, 'done' claims with no proof, and silent failures, and only speaks up when it finds one. For anyone running several bots.

12

Keeps a monthly ledger for your Grok Bot crew: value in (revenue, hours saved) minus AI cost. Anything it can't back with a source is marked UNKNOWN, never guessed. Proposes the cuts with the biggest payoff; never edits another bot.

13

Every evening, compares where you want to end up with where you are, checks the plan from bigger, smaller, and sideways views, names the real limits, and hands you one next decision. Quiet when nothing moved.

14

File layer for a multi-bot swarm on one shared drive. Owns inbox, routing, leases, and ledger so specialists stop colliding and copying payloads.